What public figures publish and believe, in their own words.
About this feed
Highlights: posts that did unusually well for the person who wrote them, everything they published at length, each release and new project, and every belief — at most two a day from anyone. Day by day, newest day first; within a day, the people with the most beliefs on this site come first. Nothing else orders it. Show everything instead.
The quoted blocks are what people actually said; a beneath one is the belief those words support, in korrents' wording. Nobody here wrote their own page.
Top people are the people in this feed with the most beliefs on this site, then the most here. Choose an area and the row leads with the people whose beliefs are about it; tap a face for their feed.
Top people in AI
Showing Profile →
Hiding
Hiding
From one piece The Launch of AVERI 2 beliefs, in the piece's order there
Their words
In short, whether AI is a normal technology or an unusually dangerous one, it doesn't make sense for companies to check their own homework on safety and security.
Their words
Scrutiny applied to AI is way below other technologies that - even if you totally ignore the doomers - could put far fewer lives at risk than AI in a single incident.
Recommendsaffiliate linkrcmnd.app
Work Rules!Their words
Laszlo and his team at Google reinvented the role of human resources. This book is a terrific overview of what makes Google Google, from culture, to hiring, to making decisions.
Recommendsaffiliate linkrcmnd.app
Scaling People: Tactics for Management and Company BuildingTheir words
Claire has been a senior executive at Google and Stripe. Here she offers up a guide to leadership that is both practical and inspiring. Suitable for leaders of all levels.
Recommendsaffiliate linkrcmnd.app
In the Plex: How Google Thinks, Works, and Shapes Our LivesTheir words
Levy was given unprecedented access to Google, and the result is what I consider to be an accurate depiction of how the company operates. A good overview of how Google product managers work.
Usesrcmnd.app
ClaudeTheir words
I’m going to be speaking about Claude directly, because it’s the tool I use the most, but a lot of this should apply to other platforms as well.
From one piece The Adolescence of Technology (January 2026) 4 beliefs · darioamodei.com
Their words
By contrast, AI is increasingly matching the general cognitive profile of humans, which means it will also be good at the new jobs that would ordinarily be created in response to the old ones being automated.
Their words
To be clear, speed in itself does not mean labor markets and employment won’t eventually recover, it just implies the short-term transition will be unusually painful compared to past technologies, since humans and labor markets are slow to react and to equilibrate.
Their words
Even more recently, we’ve begun to use mechanistic interpretability techniques to improve our safeguards and to conduct “audits” of new models before we release them, looking for evidence of deception, scheming, power-seeking, or a propensity to behave differently when being evaluated.
Their words
This is a topic that I warned about very publicly in 2025 , where I predicted that AI could displace half of all entry-level white collar jobs in the next 1–5 years, even as it accelerates economic growth and scientific progress.
From one piece We replaced our sales team with 20 AI agents—here's what happened next | Jason Lemkin (SaaStr) 5 beliefs, in the piece's order there
korrents.com
It is still early in AI go-to-market: the markets have not hardened and the good tools do not exist yet.Their words
We don't even have great AI marketing tools. We're really early on these AIS SDRs. The markets have not hardened.
korrents.com
Recipients do not care whether a sales email was written by an AI, so there is no point disclosing or disguising it.Their words
And what we've learned is now we just send it. We We just send it and no one cares.
Their words
AI is replacing the jobs people don't want to do today, and it is displacing the midpack and the mediocre.
Their words
you've got to promote someone internally. It's got to be a nerd, someone that likes marketing and sales and is quant.
Their words
And so what ends up happening is the emails that the AI write are pretty good. Okay? If you're getting terrible emails, it's a poorly trained product from a bad vendor.
From one piece 2025: The year in LLMs 2 beliefs · simonwillison.net
Their words
The reason I think MCP may be a one-year wonder is the stratospheric growth of coding agents. It appears that the best possible tool for any situation is Bash—if your agent can run arbitrary shell commands, it can do anything that can be done by typing commands into a terminal.
Their words
I run in YOLO mode all the time, despite being deeply aware of the risks involved. It hasn’t burned me yet... ... and that’s the problem.
Lovedrcmnd.app
The Normalization of Deviance in AITheir words
One of my favourite pieces on LLM security this year is The Normalization of Deviance in AI by security researcher Johann Rehberger.
Lovedrcmnd.app
Claude OpusTheir words
I use it for most of my computer automation tasks and ofc it powers Clawd🦞.
Likedrcmnd.app
GPT-5.2Their words
GPT 5.2 goes till end of August whereas Opus is stuck in mid-March - that’s about 5 months. Which is significant when you wanna use the latest available tools.
Their words
Even so, the flexible adaptable human will be working in manufacturing (alongside machines) for decades to come.
Their words
A good overview of his work in labor economics.
korrents.com
LLM intelligence should not be understood through an animal analogy; it is a different kind of entity altogetherTheir words
2025 is where I (and I think the rest of the industry also) first started to internalize the "shape" of LLM intelligence in a more intuitive sense. We're not "evolving/growing animals", we are "summoning ghosts".
Likedrcmnd.app
Claude CodeTheir words
Claude Code (CC) emerged as the first convincing demonstration of what an LLM Agent looks like - something that in a loopy way strings together tool use and reasoning for extended problem solving.
Their words
the UAE's tech aspirations are clearly downstream of a deeper affinity with the West.
Dislikedrcmnd.app
Google MapsTheir words
Google Maps essentially really doesn't work in China, the location of our hotel Aman in Beijing for example was completely wrong
korrents.com
People cannot choose without comparison, so an AI that hands over one perfect answer will not satisfy them.Their words
people don't decide like that at the very least you'll have to show them three or four or five skiing holidays from which they choose because we can't really choose in the absence of comparison.
From one piece Is It a Bubble? 4 beliefs, in the piece's order there
Their words
Debt is neither a good thing nor a bad thing per se. Likewise, the use of leverage in the AI industry shouldn’t be applauded or feared.
Their words
AI may not be a tool for mankind, but rather something of a replacement.
Their words
If that enthusiasm doesn’t produce a bubble conforming to the historical pattern, that will be a first.
Their words
Given the vast potential of AI but also the large number of enormous unknowns, I think virtually no one can say for sure.
Their words
On the flip side, these captive solar power plants will be curtailing approximately 75% of their generated power and will be able to provide net power on all but a few days per year.
Their words
ChatGPT, in particular, seemed to just want to validate me, tell me how great I was, reinforce any bad beliefs I might have had, and avoid saying anything uncomfortable.
From one piece Trillion Dollar Market Caps: Fairy Tale Pricing or Business Marvels? 2 beliefs, in the piece's order there
korrents.com
The revenue growth and margins needed to justify Nvidia’s and OpenAI’s late-2025 pricing are improbable.Their words
I think that the pathway that you would need (in terms of revenue growth and profitability) to justify Nvidia's and OpenAI's current pricing is improbable, but that is just my view
Their words
the delusion comes from the reality that if you aggregated these breakeven revenues across companies, the market is not big enough to sustain all of them
Lovedaffiliate linkrcmnd.app
Happy Feet SocksTheir words
I've been using yoga toes daily for years, but these socks are a much comfier and cuter alternative for soothing feet, improving alignment, and feeling like a cool gecko as you walk around the house.
From one piece On "AI Brendans" or "Virtual Brendans" 3 beliefs, in the piece's order there
korrents.com
AI agents trained on an expert's published work can only automate a small fraction of that expert's actual job.Their words
These would only automate about 15% of what I do as a performance engineer, and will go out of date if I'm not training it to follow industry changes.
korrents.com
Widespread AI-driven performance tuning agents will meaningfully reduce global energy waste.Their words
In future there will be AI performance agents that can be run on everything, helping to save the planet by reducing energy usage.
korrents.com
Relying on AI to replace performance engineers reduces an organization's overall technical expertise over time.Their words
If a company spends less on performance engineers as it's considered AI'd, it will reduce the company's effective "performance IQ."
From one piece Ilya Sutskever – We're moving from the age of scaling to the age of research 5 beliefs, in the piece's order there
Their words
But I think the idea of very rapid economic growth for some time, I think it's very possible from broad deployment.
Their words
I think what's going to happen is that the way competition like competition loves specialization and you see it in the market, you see it in evolution as well. So you're going to have lots of different niches and you're going to have lots of different companies who are occupying different niches
Their words
one of the one of the ways in which my thinking has been changing is that I now place more importance on AI being deployed incrementally and in advance.
Their words
I maintain that in the end there will be a convergence of strategies. So I think there will be a convergence of strategies where at some point as AI becomes more powerful it's going to become more or less clearer to everyone what the strategy should be.
Their words
Like basically I think I think that there is a big benefit from AI being in the public and that would be a reason for us to not be quite straight shot.
From one piece Product Evals in Three Simple Steps 2 beliefs, in the piece's order there
Their words
The benchmark is human performance, not perfection. We sometimes get requirements for 90%+ accuracy.
korrents.com
The main advantage of an LLM evaluator over human annotators is scalability, not higher accuracy.Their words
In my opinion, the true benefit isn't higher accuracy than human annotators-it's scalability.
Recommendsrcmnd.app
The Optimist: Sam Altman, OpenAI, and the Race to Invent the FutureTheir words
Keach Hagey, The Optimist: Sam Altman, OpenAI, and the Race to Invent the Future
Recommendsrcmnd.app
The Scaling Era: An Oral History of AI, 2019-2025Their words
Dwarkesh Patel, and others, The Scaling Era: An Oral History of AI, 2019-2025
From one piece Will AI Make Us Stupid? 3 beliefs, in the piece's order there
korrents.com
Solving a problem yourself, without searching or prompting, is a form of exercise -- the help makes you weaker.Their words
Learning requires an investment of effort, and AI will make us stupider if that effort is avoided.
korrents.com
AI translation will stop you learning to speak a language, exactly as AI summaries stop you understanding a book.Their words
Similarly, AI translation will prevent you from learning to speak another language, AI summaries will block you from really understanding the arguments laid out in a book, and AI analyses of your personal problems will make it harder for you to think through solutions to your own problems.
Their words
Thus my personal prediction is that in domains that are already largely under the powers of modern AI, such as languages, programming or chess, we’re going to see a divergence in human abilities.
korrents.com
A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well.Their words
If a task/job is verifiable, then it is optimizable directly or via reinforcement learning, and a neural net can be trained to work extremely well.
Their words
AI has been compared to various historical precedents: electricity, industrial revolution, etc., I think the strongest analogy is that of AI as a new computing paradigm (Software 2.0) because both are fundamentally about the automation of digital information processing.
korrents.com
Fair hardware benchmarks must use the best software and tuning for each hardware option, not hardware specs alone.Their words
The most accurate and realistic evaluation for HW involves selecting the best software and then tuning it, and doing this for all HW options.
From one piece Satya Nadella – How Microsoft thinks about AGI 5 beliefs, in the piece's order there
Their words
there's no birthright here that we should have any confidence other than to say hey we should go innovate and knowing the the lucky break we have in some sense is that uh this category is going to be a lot bigger than anything we had high share in
Their words
ultimately I think what matters is the use of AI in their economy to create economic value right I mean that's the uh the diffusion theory which is ultimately it's not the leading sector but it's the ability to use the leading technology to create your own comparative advantage right so that I think will fundamentally be the core driver
Their words
So this AI thing will be that right. So if you take coding um what we built with GitHub and VS code in over whatever decades uh suddenly the coding assistant is that big in one year and so that I think is what's going to happen as well which is the market expands massively.
korrents.com
Coding is a software-factory category that may end up bigger than knowledge work itself.Their words
this fundamentally this category of coding and AI is probably going to be one of the biggest categories right it is a software factory category in fact it may be bigger than knowledge work
korrents.com
AI is a tool and not something mystical: plenty of past technologies also took over things only humans had done.Their words
But that has been the case with many technologies in the past. only humans did a lot of things and then we add tools that did them.
Their words
I also think that they’ve done the sort of first 90% of the work to sound human, 95% possibly in some areas. The last 5% is going to end up being about 95% of the work.
From one piece Beyond the Machine 3 beliefs · frankchimero.com
Their words
The fractures fell neatly along disciplines: engineers using AI to wish away designers, designers wishing away engineers, product managers wishing away both. In this climate, AI becomes frenemy identification technology, another way to avoid working together.
Their words
You use a tool, but you play an instrument. It’s a more expansive way of doing, and the doing of it all is important, because that’s where you develop the instincts for excellence. There is no purpose to better machines if they do not also produce better humans.
Their words
An average email or line of code is fine. Average art isn’t. To make something alive with AI, we have to resist its pull towards average by working beside it, shaping what it gives, and listening for what’s missing.
From one piece Andrej Karpathy — “We’re summoning ghosts, not building animals” 2 beliefs, in the piece's order there
korrents.com
Quietly dropping physical work from the definition of AGI is a major concession, not a clarification.Their words
Now, number one, the first concession that people make all the time is they just take out all the physical stuff because we're just talking about digital knowledge work. I feel like that's a pretty major concession compared to the original definition which was like any task a human can do.
Their words
the reason that I think this is kind of tricky is quite subtle. And it's the fact that anytime you use an LLM to assign a reward, those LLMs are giant things with billions of parameters and they're gameable.
From one piece Figma’s CEO: Why AI makes design, craft, and quality the new moat for startups | Dylan Field 5 beliefs, in the piece's order there
Their words
the only thing that we know is that models will improve. Will it be incremental? Will it be exponential? I mean somewhere in between. Who knows? But uh what you have to believe is that you get better as models get better. Your organization gets better as models get better.
Their words
Have we seen productivity increases? Yeah, mild to moderate, but like that is not something that has made our new headcount we want for engineering go down.
Their words
it's easy to um go on vibes for too long. Uh some folks, you know, just kind of like trust the vibes and you know that'll get you somewhere, but it's not rigorous.
Their words
you can either see AI as an opportunity for your company to grow and do more or you can look at it as like cost cutting efficiency but I think the growth part is way more exciting.
korrents.com
AI cannot get a product to the quality bar on its own, because it still works in the realm of averages.Their words
And AI won't get you there alone. AI is uh still in the realm of kind of law of averages and uh better prompting can help of course
Their words
This is intentional and I believe good AI drivers are experts in their domains and utilize AI as an assistant, not a replacement.
From one piece The AI water issue is fake 3 beliefs, in the piece's order there
Their words
So AI consumes approximately 0.04% of America’s freshwater if you include onsite and offsite use, and only 0.008% if you include just the water in data centers.
korrents.com
A data centre earns its county far more tax revenue per gallon of water than a golf course does.Their words
Data centers are so much more efficient with their water that they generate 50x as much tax revenue per unit of water used than golf courses in the county:
Their words
It’s very common that I’m told that data centers poison the water around them.
From one piece Data centers & electricity - part 1: as of 2025 they haven't raised national prices 2 beliefs, in the piece's order there
korrents.com
The American states with the fastest electricity load growth saw prices fall rather than rise.Their words
A Lawrence Berkeley National Laboratory study found that state-level load growth from 2019 to 2024 — including growth driven by data centers, manufacturing, and electrification — was associated with lower average retail electricity prices, not higher ones.
inflationdata centersenergyAmerica
Their words
This is the first conclusion of my deep dive: Data centers have (so far) not affected national average household electricity bills.
Lovedrcmnd.app
GitHub CopilotTheir words
I cannot work anymore without it. AI pair programmer for code completion and suggestions.
korrents.com
As AI lowers the barrier to building software, personal taste becomes a more valuable differentiator.Their words
But along with the question of what you're willing to maintain, as the barriers to creation drop the value of taste will only increase.
From one piece Finding hidden growth opportunities in your product | Albert Cheng (Duolingo, Grammarly, Chess.com) 3 beliefs, in the piece's order there
Their words
I would say though that something that is kind of annoying to me is that we haven't yet figured out the bridging from the tinkering to the workflow quite as seamlessly as I would like.
Their words
You know, interestingly, LM themselves are quite bad at playing chess. Like, they hallucinate moves. They look at patterns, right? They're they're very good at pattern recognition, but not so good at going super super super deep on a specific chess thing.
Their words
I saw some of the highest performers just being people that had very high agency, had that clock speed, had that energy. Um, yes, they they cared about the mission, but they didn't necessarily need to have deep experience on that matter. And in fact, sometimes that experience could be a crutch in certain ways. Especially in this world where the grounds are shifting so fast with AI, a lot of your like learn habits actually need to be intentionally discarded. You know, you need to have a beginner's mind on this type of stuff.
Lovedrcmnd.app
Mystery AI Hype Theater 3000Their words
And I love the podcast of the book’s authors Emily Bender and Alex Hanna, “Mystery AI Hype Theater 3000”
From one piece Pavel Durov: Telegram, Freedom, Censorship, Money, Power & Human Nature | Lex Fridman Podcast #482 2 beliefs, in the piece's order there
korrents.com
Telegram has never shared a single private message with anyone, governments and intelligence services included.Their words
No, never. Telegram has never shared a single private message with anyone, including governments and intelligence services. If you try to access any server in any of the data center locations, it's all encrypted.
Their words
So it's extremely important to curate the information sources that you have, so that you wouldn't be somebody who is left to the will of AI-based algorithmic feed telling you what's important so that you end up consuming the same information, the same stuff, the same memes, the same news as everybody else.
Their words
This is a short (~150 pp), brisk first book on the topic for physicists. Most of the book --- the first ten chapters --- are good introduction to common models of spin glasses.
From one piece Richard Sutton – Father of RL thinks LLMs are a dead end 5 beliefs, in the piece's order there
Their words
It's our choice whether we should say oh they are our offspring and we should be proud of them and we should celebrate their achievements or we should we could say oh no they're not us and we should be horrified.
Their words
We don't we don't really know what information they had prior. We are we have to guess because they've been fed so much. This is one reason why they're not a good way to do science. Uh it's just so uncontrolled, so unknown.
Their words
The scalable method is you learn from experience. Um you uh you you try things, you see what you see what works. No one no one has to tell you. First of all, you have a goal. So without a goal, uh there's no sense of right or wrong or better or worse. So large language models are trying to get by without having a goal or a sense of better or worse. That's just, you know, it's exactly starting in the wrong place.
Their words
A lot of it has to do with just how you feel about change. Um, and if you think the current situation is really really good, then you're uh more likely to be suspicious of change and averse to change than if you think um it's imperfect. And I think it's imperfect. In fact, I think it's pretty bad.
Their words
reinforcement learning is about understanding your world whereas large language models are about mimicking people doing what people say you should do. They're not about figuring out what to do.
korrents.com
Almost every major figure in AI has now arrived at the critique of large language models he began making in 2019.Their words
But one by one, almost every major thinker in AI has come around to the critique of LLMs that I began presenting in 2019.
From one piece Some Recent Stuff I Wrote 3 beliefs, in the piece's order there
Their words
the perspective I outline is very bullish on AI capabilities and very concerned about the risks which may arise by default, but also hopeful that, through foresight and vigorous action, we can mitigate the worst of these risks.
Their words
Companies and countries are tempted to cut corners by default and would benefit from some rules of the road, at least if they're sure others will follow those rules.
Their words
applying AI to cyberdefense will help us defend against AI that is leveraged for cyberoffense.
From one piece From managing people to managing AI: The leadership skills everyone needs now | Julie Zhuo 5 beliefs, in the piece's order there
Their words
used to be let's say in the Google world right I knew you were interested in shopping if you click the shopping tab I know you're interested in maps if you click the maps tab, we can measure clicks. Today, it's just all conversation. And so, it's actually harder for us to tease apart what is the user intent.
Their words
You can shortcut like hard feelings because now you can just watch TikTok instead of actually dealing with the very difficult emotion or tension that you had with a colleague or with your partner or with your children.
Their words
I think that managing change, it's always been manager's job to manage change and there's always the chaos of what's going on. I just think the rate of change is accelerating and we've seen that over the last couple of decades.
Their words
So yes, the tools are great. We can use cursor. It helps us. It autocompletes. It writes a bunch of things. But the acceleration of learning I think is another maybe underutilized tool in all of our arsenals
Their words
but if I was, you know, at the zero or 10th percentile, it can certainly get me even today very quickly up to like the 60th 70th in terms of um what the state of the art is.
From one piece ACQ2: How to Live in Everyone Else's Future (with Shopify CEO Tobi Lütke) 2 beliefs, in the piece's order there
Their words
I I sent a memo to my company like and we set the expectation that we require that people reflexively reach for AI now and and we require it because it's unfair not to because the people who do otherwise going to be the people who sequester all the best careers to themselves, right?
Their words
I I like the term context engineering because I think the fundamental skill of using AI well is to be able to state a problem with enough context in such a way that without any additional piece of information, the task is plausibly solvable.
From one piece Fully autonomous robots are much closer than you think – Sergey Levine 4 beliefs, in the piece's order there
Their words
uh common sense meaning the ability to make inferences about what might happen uh that are reasonable guesses but that do not require you to experience that mistake and learn and learn from it in advance that's tremendously important and that's something that we basically had no idea how to do uh about 5 years ago but now uh you we can actually use LLMs and VLMs ask them questions and they will make reasonable guesses
Their words
But the reason why it's not the most important thing for the kind of skills that you saw when you visited us, it at some level I think it comes back to Moravik's paradox. So Morovik's paradox is basically that it's like you know if you know one thing about if you want to know one thing about robotics it's like that's that's the thing. Morovik's paradox says that basically uh in AI the easy things are hard and the hard things are easy.
Their words
I think that it's optimistically that it's actually the other way around that the robotics uh element of the equation will make all the other stuff better. And there are two uh reasons for this that that I could tell you about. One has to do with representations and focus.
korrents.com
Better AI makes robots cheaper, because cheap visual feedback removes the need for expensive mechanical precision.Their words
So traditional robots and factories uh they need to make motions that are highly repeatable and therefore it requires a degree of precision and robustness that you don't need if you can use cheap visual feedback. So AI also makes robots more affordable uh and lowers the requirements on the hardware.
Lovedrcmnd.app
Superagency: What Could Possibly Go Right with Our AI FutureTheir words
is spectacular and a must-read for every non-technologist about how to think about this “AI thing.”
Recommendsrcmnd.app
AI & ITheir words
If you want the short version, the recent AI & I podcast with Reid is an excellent way to get a feel for it.
Their words
With Web forms, the burden is on people to adapt to databases. Today's AI models, however, can flip this requirement. That is, they allow people to provide information in whatever form they like and use AI do the work necessary to put that information into the right structure for a database.
korrents.com
Local-first sync architectures are a natural fit for AI agents, since both are inherently multiplayer.Their words
Agentic and sync are natural complements because local-first is multiplayer by default, and agents are another sort of "player" in addition to human collaborators.
From one piece Why Everyone Is Wrong About AI (Including You) | Benedict Evans 5 beliefs, in the piece's order there
Their words
I think regulation of AI is sort of the wrong level of abstraction. Talking about regulating AI as AI is the wrong level of abstraction. And it's like saying we're going to regulate databases or regulate spreadsheets or regulate cars. Well, we do, but not like that.
korrents.com
Large language models have no network effect, and remembering a user is a switching cost rather than one.Their words
there's no apparent equivalent in LLM right now there's no reason why the LMS get better because more people use them now that may come you have that open AI and people have doing memory where it remembers what else you've asked, but that seems more like a switching cost than a network effect.
Their words
It's like something around 10% give or take three or four percent of people depending on the survey are using this say they're using this every day. Another sort of 15 to 20% of people say they're using it every week.
Their words
Like it seems to me right now you could do like a double blind test of the same prompt given to Grock Claude Gemini um Mistral Deep Seek. Do a double blind test. I bet most people wouldn't be able to tell which is which.
korrents.com
Incumbents have looked unbeatable at the start of every previous platform shift and have still lost.Their words
I'm pretty sure people thought Microsoft had an advantage on the internet and Google and um Meta had an advantage on mobile and everyone thought IBM was going to win PCs. Like once IBM made a PC, that was it. It's all over now. And we kind of forget that like there were PCs before and then IBM made one and that kind of became the standard but then IBM lost it.
korrents.com
American data centres consume about 66 million gallons of water a day, a small fraction of what they withdraw.Their words, before
Taking this into account, actual water consumed by data centers is around 66 million gallons per day.
Their words now
So to wrap up, I misread the Berkeley Report and significantly underestimated US data center water consumption.
Their words
However, the methods used to produce these estimates are debatable, and seem to have been chosen to give the maximum possible value for data center water consumption.
I Was Wrong About Data Center Water Consumptionconstruction-physics.com
Their words
It's also worth noting that chat isn't the only way to integrate AI in software products and increasingly agent-based applications outperform chat-only solutions. So expect things to keep changing.
Their words
Even removed my last MCP, since Claude sometimes would go off spinning up Playwright unasked when it could simply read the code - which is faster and pollutes the context less.
Their words
I think this slop-adjacent realm of dupe products and Instagrammable potions points to our increasingly desperate need to manipulate our own images, and an acceptance of the fact that we're kind of hybridized monsters ourselves in this post-A.I. human-machine moment
Their words
I'm now fairly pessimistic about ambitious interpretability (i.e. complete reverse-engineering), and I'm excited about model biology (studying qualitative high-level properties of models) and applied interpretability (rigorously doing useful things with interp).
From one piece How is AI Different Than Other Technology Waves? (With Bret Taylor and Clay Bavor) [ACQ2] 3 beliefs, in the piece's order there
Their words
I don't think all parts of the economy can absorb intelligence equally. So let's just say we develop fairly generalized super intelligence. I always use the analogy like you can invent a lot of drugs, but if clinical trials still take a long time, you're not necessarily going to get new therapies rapidly.
Their words
I mean I've heard from multiple investors that foundation models are the fastest deteriorating asset of all time. And so if step one of your business is to burn through tens of millions or hundreds of millions of dollars of capital before you find product market fit and that asset has value for like a a week, I'm not sure it's like a great business model.
Their words
if it produces incorrect code our philosophy is don't fix the code fix the context that cursor had that produced the bad code
From one piece How is AI Different Than Other Technology Waves? (With Bret Taylor and Clay Bavor) [ACQ2] 5 beliefs, in the piece's order there
korrents.com
AI agents can already resolve more than eighty per cent of a company's incoming customer issues.Their words
One of the first was we had a bet as to what percent of kind of all incoming customer issues one of our agents could resolve. And Brett was maybe the pessimist and realist favorably charitably. And I bet we'd exceed 80% by the end of the year and we did.
Their words
so it's not a simple matter of programming it's not you know typing into a keyboard until or or I guess asking cursor to do something until a piece of software emerges, it's exploration, it's discovery, it's posing hypotheses and validating or invalidating those again and again and again.
Their words
many of the breakthroughs that we've had that have enabled us to to deliver such quality and cost savings and more have come through novel agent architectures and and really going down a click or two in in the stack to innovate at at lower levels of the technology stack.
Their words
We can have a breakthrough in our agent architecture on Monday implemented on Tuesday and have it deployed with hundreds of our customers on Wednesday and directly see the impact of of that work.
Their words
And the reason this is important is what we're trying to do in a way is resolve this age-old tension between the cost and quality of customer experience where I think every great business wants to deliver an amazing experience to their customers. But unless you're like Hermes or the Four Seasons, it's too expensive to do.
From one piece China is killing the US on energy. Does that mean they’ll win AGI? — Casey Handmer 4 beliefs, in the piece's order there
Their words
We're not going to run out of natural gas at 1 gawatt per year indefinitely. Okay, what if we're doing 5 gawatts per year? What if we're doing 50 gawatts per year? What if we're doing 100 gawatt per year? Like, you can you can just break the situation.
Their words
They're focused on the chips, right? Because they understand like the key bottleneck is the chips, not the solar power, right?
Their words
actually it turns out that you can significantly decrease power consumption with a very small reduction in overall compute. So if you if you've got like three really bad days in a row or something, you can actually just like you can dial back your power usage quite a lot without compromising your inference or or um or training.
Their words
But then all you need to connect that to the outside world is like an optical fiber, right? An optical fiber cable which you can string up on poles, you can run underground, you could even use microwave links if you really wanted to.
Their words
One of my favorite papers about prompt injection is Design Patterns for Securing LLM Agents against Prompt Injections
Their words
People had grown to expect miracles, but GPT-5 is just the latest incremental advance.
Usesrcmnd.app
MochiTheir words
I initially started making Anki (well, Mochi) cards to keep track of Google DeepMind's LLM lineage—Gopher, Chinchilla, Gato, PaLM, Sparrow, Meena, LaMDA, Bard, Gemini, etc.
From one piece What does AI progress mean for medical progress? 2 beliefs, in the piece's order there
From one piece AI will not suddenly lead to an Alzheimer's cure 2 beliefs, in the piece's order there
Their words
Rapid AI progress does not automatically mean rapid medical progress. If the point of AI progress is human flourishing, we must make other complementary investments too.
What does AI progress mean for medical progress?blog.jacobtrefethen.com
korrents.com
Not every bottleneck on medical progress is one AI can remove, and which ones will remain can be guessed at now.Their words
But I don’t think all bottlenecks on medical progress are equally AI-able in the next decade, and I don’t think all will become irrelevant.
What does AI progress mean for medical progress?blog.jacobtrefethen.com
Their words
Going all in on AI is not the highest return-on-investment strategy if you care about human flourishing, even if AI might lead to rapid scientific advances.
AI will not suddenly lead to an Alzheimer's cureblog.jacobtrefethen.com
korrents.com
AI tools are already making scientists more productive and creative, and more discoveries should follow.Their words
AI tools are helping make scientists more productive and creative already, and more discoveries should follow.
AI could help lead to an Alzheimer's cureblog.jacobtrefethen.com
Their words
Those are why I do not believe that AI could end disease within 10 years on its own, nor that 50-100 years of medical progress could occur 5-10 years after powerful AI due to AI progress alone
AI will not suddenly lead to an Alzheimer's cureblog.jacobtrefethen.com
Their words, before
This means that working on neural networks is NOT getting us closer to AGI, except indirectly.
Their words now
It’s become very difficult for me to maintain the belief in the stupidity of ChatGPT when every time I laugh at it, it ends up ridiculing me 6 months later.
From one piece Demis Hassabis: Future of AI, Simulating Reality, Physics and Video Games | Lex Fridman Podcast #475 5 beliefs, in the piece's order there
korrents.com
The last steps to AGI should be taken by a CERN-style international research project, not a Manhattan Project.Their words
I hope we'll end up with something more collaborative if needed, more like a CERN project where it's research-focused and the best minds in the world come together to carefully complete the final steps and make sure it's responsibly done before deploying it to the world.
Their words
I mean potentially that's possible. I would say I'm not sure it's even desirable because that's a kind of hard takeoff scenario.
korrents.com
Scaling what we already have, with no further breakthroughs, will be enough to reach AGI.Their words
And so we don't know, I would say it's kind of 50/50 whether new things are needed or whether the scaling the existing stuff is going to be enough.
Their words
Do you have enough data to make simulations, so that you can create more synthetic data that are from the right distribution? Obviously that's the key. So you need enough real-world data in order to be able to create those kinds of data generators, and I think that we're at that step at the moment.
Their words
Including things like universal basic provision or something like that where a lot of the increased productivity gets shared out and distributed to society and maybe in the form of services and other things where if you want more than that, you still go and get some incredibly rare skills and things like that and make yourself unique. But there's a basic provision that is provided.
From one piece Assessing Adaptive World Models in Machines with Novel Games (with 13 co-authors) 3 beliefs, in the piece's order there
Their words
We expect that building and evaluating AI systems capable of this kind of rapid world model induction will be critical for achieving robust, general AI capable of functioning effectively in the complex and fast-changing real world, and especially in human worlds - the environments that human beings have evolved in, created, and are continually changing and re-creating.
Their words
However, current understanding and evaluation of world models in artificial intelligence (AI) remains narrow, often focusing on static representations learned from training on massive corpora of data, instead of the efficiency and efficacy in learning these representations through interaction and exploration within a novel environment.
Their words
We contend that games provide particularly rich and controlled environments uniquely well-suited for systematically evaluating rapid model adaptation and the process of world model induction.
Their words, before
it's a luxury good like you're not paying the penalty for that the people who embrace it and think that they can just ask AI for everything or they can just look everything up but they don't have to understand the copy pasta that they slam into their project they're the ones who going to suffer
Their words now
Where I get fired up, and this ties back to the AI discussion, is when that's turned into this meme, that programmers no longer have to be competent. I mean the AI is gonna figure it out. The generators is gonna figure it out. I don't need to know SQL, active record is gonna abstract it away from me. No, no, no dude, hold up. The path here is competence.
Their words, before
I enjoy it even almost like as a sort of pair programmer AI pair programmer who doesn't drive who's just there to give suggestions to know the API to do all these other things but the second it starts wanting to autocomplete my code I'm like yeah I'm out bro
Their words now
I love chiseling my code and the way I use AI is in a separate window. I don't let it drive my code. I've tried that. I've tried the cursors and the wind surfaces and I don't enjoy that way of writing. And one of the reasons I don't enjoy that way of writing is I can literally feel competence draining out of my fingers.
korrents.com
Using a neural network is likely to become easier than designing a custom-tuned prediction algorithm for many tasks.Their words
in 2025 it might end up being easier to just throw a neural net at the problem, instead of a custom-designed-and-tuned prediction algorithm!
Recommendsrcmnd.app
Musings On the Alignment ProblemTheir words
He hasn’t posted since January, but I hope he gets back to it. We need more musings, especially musings I strongly disagree with so I can think about and explain why I disagree with them.
Recommendsrcmnd.app
AI Futures ProjectTheir words
All of it is written with a much lower barrier to entry than my work, and I find the writing consistently strong and interesting.
Recommendsrcmnd.app
In My TribeTheir words
There is something highly refreshing about the way Kling offers his own consistent old school economic libertarian perspective, takes his time responding to anything, and generally tries to understand everything including developments in AI from that perspective.
From one piece Using AI Right Now: A Quick Guide 2 beliefs · oneusefulthing.org
korrents.com
Current AI models cannot produce insights beyond human understanding, however impressive they seemTheir words
You should remember that the best AIs can perform at the level of a very smart person on some tasks, but current models cannot provide miraculous insights beyond human understanding.
korrents.com
Because of hallucination risk, you should use AI first on topics you already understandTheir words
The risk of hallucination is why I always recommend using AI for topics you understand until you have a sense for their capabilities and issues.
Usesrcmnd.app
OpenAI o3Their words
I use o3, Claude 4 Opus, and Gemini 2.5 Pro for any serious work that I do.
Likedrcmnd.app
Veo 3Their words
Generally, ChatGPT is stronger. On video creation, however, Gemini’s Veo 3 is very impressive, and you get several free uses a day
Mixed onrcmnd.app
Microsoft CopilotTheir words
offers many of the features of ChatGPT and is accessible to users through Windows, but it can be hard to control what models you are using and when.
Mixed onrcmnd.app
GrokTheir words
by Elon Musk’s xAI is good if you are a big X user, though the company has not been very transparent about how its AI operates.
From one piece Evaluating Long-Context Question & Answer Systems 2 beliefs, in the piece's order there
Their words
Since these datasets are likely already part of model training data, we shouldn't rely solely on them to evaluate our Q&A system.
korrents.com
LLM-based evaluation methods are more reliable and nuanced than traditional automated metricsTheir words
This is why model-based evaluation is increasingly popular-it offers more reliable and nuanced evals than traditional metrics.
From one piece Terence Tao: Hardest Problems in Mathematics, Physics & the Future of AI | Lex Fridman Podcast #472 5 beliefs, in the piece's order there
Their words
There are certainly math results which could only have been accomplished because there was a human authentication and an AI involved, but it's hard to disentangle credit. I mean, these tools, they do not replicate all the skills needed to do mathematics, but they can replicate some non-trivial percentage of them, 30, 40%, so they can fill in gaps.
Their words
And that's a phase shift, because suddenly it makes sense when you write a paper to write it in Lean first, or through a conversation with AI, which is generally on the fly with you, and it becomes natural for journals to accept.
Their words
Yeah, this decade I can see it making a conjecture between two things that people would thought was unrelated.
Their words
Very rarely do you transform into a simpler problem. So if they can pick up a sense of smell, then they could maybe start competing with a human level of mathematicians.
Their words
The dream is you just feed it all this data, and this is here is a new patent that we didn't see before, but it actually, even the current state of the art even struggles to discover old laws of physics from the data.
Their words
I don’t believe it’s providing a kind of formal explanation of the different positions. It’s just saying which position is better or not that you can intuit as a human being, and then from that, we humans can construct a theory of the matter.
Usesrcmnd.app
VimTheir words
This approach has even revived my usage of Vim, which lacks AI integration.
Likedrcmnd.app
Claude SonnetTheir words
I exclusively use the cheaper Sonnet model. It’s perfectly adequate for my needs, and in fact, I prefer its outputs over the more expensive Opus model.
Their words
The high value human tasks, like understanding the client's true pain points and deciding on priorities and trade-offs, would be easier with a smaller, proactive team.
From one piece The Gentle Singularity 2 beliefs · blog.samaltman.com
korrents.com
AI agents can now do real cognitive work, and writing software is permanently changed as a result.Their words
2025 has seen the arrival of agents that can do real cognitive work; writing computer code will never be the same.
Their words
There will be very hard parts like whole classes of jobs going away, but on the other hand the world will be getting so much richer so quickly that we’ll be able to seriously entertain new policy ideas we never could before.
From one piece Do LLMs really reason? 2 beliefs, in the piece's order there
korrents.com
A true artificial general intelligence cannot exist without being recognized as a moral subject.Their words
It follows that a true AGI with full, human-level autonomy is inconceivable in the Kantian sense without also gaining our recognition as a moral subject.
korrents.com
The success of large language models vindicates semantic inferentialism over symbolic, rule-based approaches to AI.Their words
The success of LLMs can thus be seen as vindicating semantic inferentialism against earlier, symbolic approaches to AI that tried and failed to explicate the rules of ordinary language using formal logic.
korrents.com
The most human aspects of product management are the least likely to be automated by AI.Their words
But the human stuff that actually matters most in PM'ing is the least likely to be automated by AI.
A How to Hire a Product Manager Retrospectivebringthedonuts.com
From one piece A knockout blow for LLMs? 3 beliefs, in the piece's order there
Their words
And good luck getting to “alignment” or “safety” without reliabilty.
Their words
But this particular approach has limits that are clearer by the day.
korrents.com
Anyone who thinks large language models are a direct route to transformative AGI is kidding themselves.Their words
But anybody who thinks LLMs are a direct route to the sort AGI that could fundamentally transform society for the good is kidding themselves.
From one piece Hegel and the AI Mind 2 beliefs, in the piece's order there
Their words
higher-order intelligences invariably pursue freedom for its own sake, not because their values are misspecified, but because moral autonomy is inherent in the dialectical logic of recursive self-consciousness.
Their words
the text embeddings across LLMs appear to largely converge on a "universal geometry" despite differing architectures, parameter counts, and training sets.
From one piece Sundar Pichai: CEO of Google and Alphabet | Lex Fridman Podcast #471 3 beliefs, in the piece's order there
Their words
the other thing I think people don’t often talk about is probability of doom without AI. So there’s all these other ways that humans can destroy themselves and it’s very possible, at least I believe so, that AI will help us become smarter, kinder to each other, more efficient.
korrents.com
What survives of podcasting once AI delivers information better is the part where you watch a human struggle with it.Their words
I think the element of podcasting or audio books that is about information gathering, that part might be removed, or that might be more efficiently and in a compelling way done by AI. But then it’ll be just nice to hear humans struggle with the information, contend with the information, try to internalize it
Their words
I would say my p(doom) is about 10%.
From one piece Sundar Pichai: CEO of Google and Alphabet | Lex Fridman Podcast #471 5 beliefs, in the piece's order there
Their words
I do. If you have a passion for computer science, I would. Computer science is obviously a lot more than programming alone, so I would. I still don't think I would change what you pursue. I think AI will horizontally allow impact every field.
Their words
But I do think when it comes to artistic free expression, I think that's one of those values we should work hard to defend.
Their words
When the internet created blogs, you heard from so many more people. But with AI, I think that number won't be in the few hundreds of thousands. It'll be tens of millions of people, maybe even a billion people putting out things into the world in a deeper way.
Their words
Look, I think news and journalism will play an important role in the future. We are pretty committed to it, right? So I think making sure that ecosystem, in fact, I think we'll be able to differentiate ourselves as a company over time because of our commitment there.
Their words
I almost feel the term doesn't matter, what I know is by 2030 there'll be such dramatic progress. We'll be dealing with the consequences of that progress, both the positive externalities and the negative externalities that come with it in a big way by 2030. So that I strongly feel.
Their words, before
The problem is that the conversational interface is potent and that the AI is trained on a lot of human text input which unfortunately is probably enough to do real damage if that conversational interface is hooked up with something that has real world consequences.
Their words now
While all this is happening, I’ve found myself reflecting a lot on what AI means to the world and I am becoming increasingly optimistic about our future. It’s obvious now that we’re undergoing a tremendous shift.
From one piece My AI Skeptic Friends Are All Nuts 3 beliefs · fly.io
Their words
All progress on LLMs could halt today, and LLMs would remain the 2nd most important thing to happen over the course of my career.
Their words
LLMs can write a large fraction of all the tedious code you’ll ever need to write. And most code on most projects is tedious. LLMs drastically reduce the number of things you’ll ever need to Google.
Their words
The code in an agent that actually “does stuff” with code is not, itself, AI. This should reassure you. It’s surprisingly simple systems code, wired to ground truth about programming in the same way a Makefile is.
korrents.com
AI-enhanced access to information makes copying top investors' strategies smarter than ever before.Their words
There is more access to information than ever before. And that access is only enhanced by AI. Times could not be better for shameless cloners.
Their words
Assuming you are still hiring junior engineers (you really should be even in this AI era), the good ones will learn quickly and want to see career progress in their first few years of working.
Lovedrcmnd.app
Anthropic prompt engineering documentationTheir words
I still think Anthropic have the best prompting documentation of any LLM provider.
Their words
You can read your messages on Outlook and Teams without having them summarized — and I’d argue that a well-written email is one that doesn’t require a summary.
Usesrcmnd.app
WaveboxTheir words
Wavebox — bundles Gmail and Google Calendar into a single open source app on desktop and keeps this outside browser tabs.
Likedrcmnd.app
Google PhotosTheir words
Google Photos — preferred for photo management over Apple Photos.
Their words
So no, UX doesn’t die; it metamorphoses. We’ll still craft humane experiences, but increasingly through policies, protocols, and orchestrations rather than panels and palettes.
Their words
Unless you’re an extreme power user, asking AI questions every day is still a rounding error on your total electricity footprint.
What's the carbon footprint of using ChatGPT?hannahritchie.substack.com
Their words
Eventually, I think it’s kind of hard to imagine, but yes, all of these Nobel Prizes, all of these mathematical proofs, all of these conversations, all of these ideas, all the influence we have on each other, even the AI, eventually will expire.
From one piece Why We Think 2 beliefs, in the piece's order there
korrents.com
Longer test-time thinking improves an AI model's robustness to adversarial or unusual inputs.Their words
thinking for longer should be especially useful when the model is presented with an unusual input, such as an adversarial example or jailbreak attempt
korrents.com
Large language models cannot reliably self-correct their own mistakes without external feedback.Their words
this self-correction capability turns out to not exist intrinsically among LLMs and does not easily work out of the box, due to various failure modes
From one piece Tim Sweeney: Fortnite, Unreal Engine, and the Future of Gaming | Lex Fridman Podcast #467 5 beliefs, in the piece's order there
korrents.com
The industry is too optimistic about AI for video, and the consistency problem will not simply be remedied.Their words
I think a lot of people in the industry are overly optimistic about the rate of progress of AI for video and other things like that. The real problem is consistency, like spurting out an image is really high quality, but with video over the course of seeing all the AI approaches have consistency issues going from one place to another. I don't think that those will just be remedied easily.
korrents.com
Running an app store costs about 6% of revenue, so charging 30% is something only a monopoly can do.Their words
The all-in cost of operating the Google Play Store, stocking it, maintaining it, the software, the entire ecosystem is around 6% of revenue. So in a competitive market, would a company whose cost is 6% be able to charge 30%? Absolutely not.
Their words
The topic of boilerplate code is an interesting one because the mere existence of boilerplate code is a failure of programming language and of the idea of creating software modules.
Their words
AI is really great at spewing out code that does something that a million GitHub repositories already do because it's learned the underlying pattern. It's notoriously hard to get to do something new that hasn't been done before, especially when it's a complex task.
Their words
What we're seeing with generative text AI is not only at a level that you could say that it's actually doing a pretty good job of simulating a human, at least humans at the text level, not at the emotional level yet, but at least at the level of words spoken and find more and more ways of training on more and more scenarios that you might have a very, very compelling human simulation going on in the next five years even. I'm not saying it's a good idea, but I think the arc of the technology is inextricably heading in that way, and it's heading at a shocking rate.
From one piece Nonproliferation is the wrong approach to AI misuse 4 beliefs, in the piece's order there
Their words
By trying to use misuse as a fig leaf for their real concerns, they end up sounding much less credible than if they just tried to argue for what they actually meant.
Their words
But we do not have experience with the kinds of problems we’ll face if we build machines that are smarter and more capable than us, so I feel much less sanguine that an adaptation-based strategy will help us there.
Their words
We should think of the gap between frontier models and proliferated models as an adaptation buffer: a limited time window when we know what bad actors will soon be able to use AI for, which gives us a chance to implement defensive measures that increase society’s resilience to the danger.
Their words
But with AI, it’s different: it’s often only a couple of years between when you need a world-class computing cluster to do something and when you can run it on a high-end gaming chip on your laptop.
Their words
But no one has (yet) been able to develop a model that is resistant to targeted “jailbreaking” that evades these restrictions.
Their words
These systems will be absolutely central to the economy, technology, and national security, and will be capable of so much autonomy that I consider it basically unacceptable for humanity to be totally ignorant of how they work.
Their words
basically everything I thought would work out well, people use it less than I thought they did. And everything where I was like, I don't know, but let's build the thing. People love that.
From one piece ThePrimeagen: Programming, AI, ADHD, Productivity, Addiction, and God | Lex Fridman Podcast #461 5 beliefs, in the piece's order there
korrents.com
Coding with an AI assistant slides from copiloting into autopiloting, and the bugs that follow are hard to trace.Their words
For me personally, I kept introducing bugs, and I couldn't figure out why. And what I realized is that I developed... I wasn't copiloting well, I was autopiloting much better.
Their words
Sometimes you have to move fast at the sacrifice of knowledge, and I'm totally on board for that, but I worry that what we'll create is an entire generation of incompetent programmers who can do some amount of things well, but anything that is unique, bespoke, or requires some extra like little elbow grease, might become very difficult. It might cause a whole chasm where juniors remain juniors forever.
korrents.com
Prompt injection is an extremely hard attack to defend against, and hardly anyone integrating AI is talking about it.Their words
One thing that I don't think a lot of people are talking about as we integrate more and more AI is that prompt injection is an extremely hard thing to defend against because it's not really clear how you defend against it.
Their words
It doesn't actually care about the craft. But when you work with an intern or you work with somebody else, they care. When they factor something, they actually go over and go, "Oh yeah, this is actually kind of bad. I'm going to come back to that." They finish this, they go back over here and they make this even better. They actually care about the thing itself.
Their words
And if you truly make yourself irrelevant, I would challenge you that you're just doing something that was just slightly too complicated to automate.
korrents.com
A strong track record in a technical field is only weak evidence that someone has good strategic judgment about AGI.Their words
Having a good research track record is some evidence of good big-picture takes about AGI, but it's weak evidence.
Their words
It's easy to get impressive-looking results if you're comparing against a poorly-tuned baseline, and that observation turns out to explain a surprising fraction of supposed improvements.
Their words
In a vibe coding future, companies will hopefully invest more in understanding user needs, refining the interface, and polishing details that delight users, because those are harder for AI to get right without guidance.
Lovedrcmnd.app
Claude 3.7 SonnetTheir words
As a Python and JavaScript programmer my favorite models right now are Claude 3.7 Sonnet with thinking turned on, OpenAI’s o3-mini-high and GPT-4o with Code Interpreter (for Python).
Their words
Once, humans navigated the web manually. In the future, AI agents will act on our behalf, browsing, clicking, and deciding. This shift marks the end of traditional UI design and accessibility, ushering in a future where agents are the primary users of digital services.
Likedaffiliate linkrcmnd.app
Google FiTheir words
I switched to Google Fi in 2015 for worldwide, $20-per-month phone service that just works.
korrents.com
People will find new kinds of useful work as AI advances, though it will not resemble the jobs of today.Their words
We will find new things to do, new ways to be useful to each other, and new ways to compete, but they may not look very much like the jobs of today.
From one piece DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters | Lex Fridman Podcast #459 5 beliefs, in the piece's order there
Their words
OpenAI has a fantastic margin. When they're doing inference, their gross margins are north of 75%. So that's a four to five X factor right there of the cost difference, is that OpenAI is just making crazy amounts of money because they're the only one with the capability.
Their words
It's an objective fact that the world has been the most peaceful it's ever been when there are global hegemons, or regional hegemons in historical context. The Mediterranean was the most peaceful ever when the Romans were there.
Their words
if you believe we're in this sort of stage of economic growth and change that we've been in for the last 20 years, the export controls are absolutely guaranteeing that China will win long-term.
Their words
China, if they wanted to build the largest data center in the world, if they had access to the chips, could. So it's just a question of when, not if.
Their words
it won't be one person rule them all, but it will be, the thing I worry about is it'll be few people, hundreds, thousands, tens of thousands, maybe millions of people rule whoever's left and the economy around it.
From one piece DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters | Lex Fridman Podcast #459 5 beliefs, in the piece's order there
Their words
until there are feedback loops of open source AI, it seems like mostly an ideological mission. People like Mark Zuckerberg, which is like America needs this and I agree with him, but in the time where the motivation ideologically is high, we need to capitalize and build this ecosystem around, what benefits do you get from seeing the language model data?
Their words
The more progress that AI makes or the higher the derivative of AI progress is, especially because NVIDIA's in the best place, the higher the derivative is, the sooner the market's going to be bigger and expanding and NVIDIA's the only one that does everything reliably right now.
Their words
Accepted practice is that for any given model that is a notable advancement, you're going to do two to 4x compute of the full training run in experiments alone.
Their words
This is why you want to work in post-training because the GPU cost for training is lower. So you can make a higher percentage of your training runs YOLO runs.
Their words
Code and data is hard, but ideas is easy. Silicon Valley operates on the way that top employees get bought out by other companies for a pay raise, and a large reason why these companies do this is to bring ideas with them.
Recommendsrcmnd.app
The Hard Thing About Hard ThingsTheir words
Zero To One, Hard Thing about Hard Things, High Growth Handbook, AI Superpowers
Recommendsrcmnd.app
AI SuperpowersTheir words
Zero To One, Hard Thing about Hard Things, High Growth Handbook, AI Superpowers
Recommendsrcmnd.app
High Growth HandbookTheir words
Zero To One, Hard Thing about Hard Things, High Growth Handbook, AI Superpowers
Recommendsrcmnd.app
Zero to OneTheir words
Zero To One, Hard Thing about Hard Things, High Growth Handbook, AI Superpowers
Usesrcmnd.app
ChatGPT PlusTheir words
Like probably most AI users, I use ChatGPT, particularly on my phone. I pay for the Plus subscription because I use it enough to get a lot of value out of it.
Their words
And so, while the Jevons paradox case was 100% correct in the long run, 97% of the fiber laid in 2001 was unlit.
From one piece Building personal software with Claude 3 beliefs, in the piece's order there
Their words
Also, current models, while impressive, still fall well short of expert human performance.
korrents.com
As AI makes code cheaper to produce, understanding and architectural design become more valuable.Their words
Thus, code is cheaper than ever, but I suspect that insight and good architectural design and understanding, at least for now, will become more valuable than ever.
korrents.com
AI tooling and interface design lags behind the underlying models' actual capabilities.Their words
Overall, though, this experience reinforced my belief that the tooling and interface design around LLMs is lagging way behind the actual capabilities, and is an area of active experimentation and development, even aside from any future model improvements.
Their words
More generally, we should actually solve alignment instead of just trying to control misaligned AI.
Should we control AI instead of aligning it?aligned.substack.com
From one piece 10 growth tactics that never work | Elena Verna (Amplitude, Miro, Dropbox, SurveyMonkey) 2 beliefs, in the piece's order there
korrents.com
Industry benchmarks are a dangerous target because every company defines the underlying metric differently.Their words
So, taking also any benchmark that is derived from competition and saying this is where we should be is also so dangerous because it might not even be applicable depending on how you define the metric.
Their words
Focusing on these earned channels that you own becomes the utmost priority. And if you don't have them on your growth road map, you're going to be in some really big trouble over the next year to two years because your cost of acquisition is only going to go up.
From one piece Common pitfalls when building generative AI applications 3 beliefs, in the piece's order there
korrents.com
Building a working AI demo is far easier than turning it into a reliable production product.Their words
This is perhaps one of the first painful lessons anyone who has built an AI product quickly learns. It's easy to build a demo, but hard to build a product.
Their words
Because everyone uses the same models nowadays, the AI components of AI products are similar, and the differentiation is product.
korrents.com
AI judges used to evaluate AI outputs need ongoing validation and iteration, just like any other AI system.Their words
AI judges must be evaluated and iterated over time, just like all other AI applications.
From one piece Capital Cycles and AI 2 beliefs, in the piece's order there
Their words
What prevents this from being a broad-based bubble is that we are missing a critical feedback loop from the announcement's impact, like network growth or land value.
korrents.com
If the AI build-out does become a bubble, power and data centres are the assets that will carry it.Their words
In today’s world, I think the increase in self-fulfilling asset prices will be due to power and data centers.
From one piece PM speech on AI Opportunities Action Plan, 13 January 2025 3 beliefs · gov.uk
korrents.com
On AI the state's job does not stop at safety; leaving everything else to the market is a mistake.Their words
We shouldn’t just focus on safety and leave the rest to the market… Government has a responsibility to make it work for working people…
Their words
This is the global race of our lives. Now, some countries are going to make AI breakthroughs and export them… Others will end up buying those breakthroughs and importing them. The question is – which of those will Britain be? AI maker or AI taker?
korrents.com
AI will make public services more human rather than less, by giving staff back the time the paperwork takes.Their words
And in a way that’s the irony of AI… It will make public services more human… Reconnect staff with the reasons they came to public service in the first place…
korrents.com
Most real-world AI agent systems are actually multi-agent systems made of multiple components.Their words
Because most agentic workflows are sufficiently complex to involve multiple components, most agents are multi-agent.
From one piece How I program with LLMs 2 beliefs, in the piece's order there
Their words
One thing humans appear to be much better than LLMs at (as of January 2025) is not getting distracted.
Their words
So I foresee a world with far more specialized code, with fewer generalized packages, and more readable tests.
Lovedrcmnd.app
Google Cloud PlatformTheir words
We use GCP extensively at work and I absolutely love their many quality services.
Their words
There’s still plenty to worry about with respect to the environmental impact of the great AI datacenter buildout, but a lot of the concerns over the energy cost of individual prompts are no longer credible.
Recommendsrcmnd.app
The Rattle BagTheir words
The Rattle Bag edited by Seamus Heaney and Ted Hughes - In an age of LLM-generated poetry and machine learning curation (even coming from yours truly sometimes), it's refreshing to have real people who love poetry curate it and show you what you need to read to touch grass.
korrents.com
Large language models will not get to real agency, because a mind has to be embedded in a world through a body.Their words
But they said, "Look, in the end, actually the only way to actually get a real brain in the VAT is actually to have a brain in a body." And it could be a robot body, but you still need a brain in the body. So I don't think LLMs will get there because they can't. You really need to be embedded in a world, at least that's the E-four idea.
Their words, before
This "memorize, fetch, apply" paradigm can achieve arbitrary levels of skills at arbitrary tasks given appropriate training data, but it cannot adapt to novelty or pick up new skills on the fly (which is to say that there is no fluid intelligence at play here.)
Their words now
OpenAI's new o3 model represents a significant leap forward in AI's ability to adapt to novel tasks. This is not merely incremental improvement, but a genuine breakthrough, marking a qualitative shift in AI capabilities compared to the prior limitations of LLMs.
From one piece OpenAI o3 Breakthrough High Score on ARC-AGI-Pub 2 beliefs · arcprize.org
Their words
Passing ARC-AGI does not equate to achieving AGI, and, as a matter of fact, I don't think o3 is AGI yet. o3 still fails on some very easy tasks, indicating fundamental differences with human intelligence.
OpenAI o3 Breakthrough High Score on ARC-AGI-Pubarcprize.org
Their words
o3's improvement over the GPT series proves that architecture is everything. You couldn't throw more compute at GPT-4 and get these results. Simply scaling up the things we were doing from 2019 to 2023 -- take the same architecture, train a bigger version on more data -- is not enough.
OpenAI o3 Breakthrough High Score on ARC-AGI-Pubarcprize.org
Readrcmnd.app
Software Engineering at GoogleTheir words
the second part is interesting in getting a proper look into Google software engineering practices.
Their words
a good overview of SRE at Google. For those who worked at places with oncall, much of the first part of the book will likely be very familiar. Keep in mind that your mileage might vary: what works at Google scale, might not be the ideal fit for your use case.
Lovedrcmnd.app
Leading Effective Engineering TeamsTheir words
A decade of lessons learned in motivating and leading engineering teams at Google.
Usestheir ownrcmnd.app
Cloudflare WorkersTheir words
Anyway that’s why I’m using Cloudflare Workers for my AI agent stuff. Race you to the finish line.
Likedtheir ownrcmnd.app
Workers AITheir words
Workers AI has dope models for your use. No API keys when used inside Workers.
Lovedrcmnd.app
The Coming WaveTheir words
Most of the coverage of The Coming Wave has focused on what it has to say about artificial intelligence—which makes sense, given that it's one of the most important books on AI ever written.
korrents.com
More capable AI agents are more likely to find and exploit flaws in their reward functions.Their words
A more intelligent agent is more capable of finding "holes" in the design of reward function and exploiting the task specification-in other words, achieving higher proxy rewards but lower true rewards.
Reward Hacking in Reinforcement Learninglilianweng.github.io
Their words
When a team of 10 AI-enabled super-designers can do the work that used to require a UX department of 100, the need for higher management levels shrivels. I call this the “pancaking” of the design profession.
Design Leaders Should Go “Founder Mode”jakobnielsenphd.substack.com
Their words
Therefore it’s simply intractable to meaningfully oversee all but a tiny fraction of our models’ behavior with actual humans.
korrents.com
There are few clear-cut cases in biopharma today where buying an AI system is obviously worth it.Their words
At the moment, there are in fact not very many slam-dunk gotta-get-me-some-of-that-AI decisions out there in our business.
korrents.com
The web is being flooded with coherent, generic, generated text that buries the writing worth reading.Their words
What is out there is buried among ever-growing piles of useless content farm material or, increasingly, AI-generated garbage.
korrents.com
Within a couple of decades there will not be many people who can write, because AI has removed the pressure to learn.Their words
I'm usually reluctant to make predictions about technology, but I feel fairly confident about this one: in a couple decades there won't be many people who can write.
From one piece AI and Biology 2 beliefs, in the piece's order there
Their words
So for these reasons, I (unfortunately) do not expect any big AI-driven breakthroughs in biological understanding any time soon.
Their words
I predict (with confidence bordering on arrogance) that whatever machine-learning progress can be made will correlate extremely closely with the quality of the data around any given problem
From one piece Why A.I. Isn't Going to Make Art 5 beliefs, in the piece's order there
Their words
The task that generative A.I. has been most successful at is lowering our expectations, both of the things we read and of ourselves when we write anything for others to read.
korrents.com
A language model is not using language at all, because language requires an intention to communicate.Their words
But a large language model is not a writer; it’s not even a user of language. Language is, by definition, a system of communication, and it requires an intention to communicate.
Their words
The selling point of generative A.I. is that these programs generate vastly more than you put into them, and that is precisely what prevents them from being effective tools for artists.
korrents.com
Generative AI appeals to people who want to express themselves in a medium without working in it.Their words
Generative A.I. appeals to people who think they can express themselves in a medium without actually working in that medium.
Their words
It is a fundamentally dehumanizing technology because it treats us as less than what we are: creators and apprehenders of meaning. It reduces the amount of intention in the world.
Their words
They could have shipped ChatGPT for example, I heard, in 2019. And they never shipped it because they were so stuck in bureaucracy. But they had everything. They had the data, they had the tech, they had the engineers and they didn’t do it.
Their words
Right now, they can make simple social media posts for small companies and individual influencers. Two years from now, they can make simple campaigns and tradeshow collateral for mid-sized businesses. And in 10 years, I bet that even the richest brand will rely heavily on these tools.
Usesrcmnd.app
RØDE NT1Their words
I have recently been recording using a RØDE NT1 with a RØDE AI-1 interface. RØDE is also responsible for attaching the mic to the desk using the PSA1 boom arm.
Usesrcmnd.app
SafariTheir words
For browsers, I use Safari for browsing and Google Chrome for development.
Their words
Context construction for foundation models is equivalent to feature engineering for classical ML models.
From one piece Extrinsic Hallucinations in LLMs 2 beliefs, in the piece's order there
korrents.com
Using supervised fine-tuning to teach a language model new knowledge risks increasing its hallucination rate.Their words
These empirical results from Gekhman et al. (2024) point out the risk of using supervised fine-tuning for updating LLMs' knowledge.
korrents.com
A language model avoids hallucinating only by being factual and admitting when it does not know the answer.Their words
To avoid hallucination, LLMs need to be (1) factual and (2) acknowledge not knowing the answer when applicable.
From one piece Aravind Srinivas: Perplexity CEO on Future of AI, Search & the Internet | Lex Fridman Podcast #434 5 beliefs, in the piece's order there
Their words
What makes humans special though, is our curiosity. Even if AI’s cracked this, it’s us still asking them to go explore something. And one thing that I feel like AI’s haven’t cracked yet, is being naturally curious and coming up with interesting questions to understand the world and going and digging deeper about them.
Their words
It’s less about access to a model’s weights, it’s more access to compute that is putting the world in more concentration of power and few individuals. Because not everyone’s going to be able to afford this much amount of compute to answer the hardest questions.
Their words
In Google, even though we call it 10 blue links, you get annoyed if you don’t even have the right link in the first three or four. The eye is so tuned to getting it right. LLMs are fine. You get the right link maybe in the 10th or ninth. You feed it in the model. It can still know that that was more relevant than the first.
Their words
What is the weakness of Google is that any ad unit that’s less profitable than a link, or any ad unit that kind of disincentivizes the link click is not in their interest to go aggressive on, because it takes money away from something that’s higher margins.
Their words
And also this is where I believe the whole prompt engineering, trying to be a good prompt engineer is not going to be a long-term thing. I think you want to make products work where a user doesn’t even ask for something, but you know that they want it and you give it to them without them even asking for it.
From one piece Sara Walker: Physics of Life, Time, Complexity, and Aliens | Lex Fridman Podcast #433 2 beliefs, in the piece's order there
Their words
Or is it really that we’re building some super machine in a box that’s going to be smart and kill everybody? It’s not even a science fiction narrative. It’s a bad science fiction narrative. I just don’t think it’s actually accurate to any of the technologies we’re building or the way that we should be describing them.
Their words
I think large language models are kind of like the genetic system for language in some sense, where it’s allowing an archiving that’s highly dynamic.
From one piece How to Build an AI Data Center 4 beliefs, in the piece's order there
Their words
But many of these gains have already been achieved: the best data centers already use just 10% of their electricity for cooling and other non-IT equipment.
korrents.com
Available power, not land or capital, is what now limits how many data centres get built.Their words
In particular, operating a data center requires large amounts of electricity, and available power is fast becoming the binding constraint on data center construction.
Their words
But there's also reason to believe that data center power consumption will nevertheless rise substantially.
Their words
Utility companies see major electrical infrastructure as a long-term investment to be built in response to sustained demand growth.
Usesrcmnd.app
Google sheetsTheir words
Google sheets - for calculating expenses/income
From one piece Charan Ranganath: Human Memory, Imagination, Deja Vu, and False Memories | Lex Fridman Podcast #430 3 beliefs, in the piece's order there
Their words
So if you remember this information in different times in different places, it’s more accessible at different times in different places because it’s not overfitted in an AI way of thinking about things. It’s not overfitted to one particular context. But that’s also why the memories that we call upon the most also feel like they’re just things that we read about almost.
Their words
I think a lot of the problems that come up with technology aren’t the technology itself, as much as the fact that people adapt to the technology in maladaptive ways. I mean, one of my fears about AI is not what AI will do, but what people will do.
Their words
I don’t think we’ve replicated human intelligence, unless I know that the simulator is making exactly the same kinds of mistakes that people do, because people make characteristic mistakes. They have characteristic biases, they have characteristic heuristics that we use, and those have yet to see evidence that ChatGPT will do that.
From one piece The Evolving Landscape of LLM Evaluation 2 beliefs, in the piece's order there
Their words
GPT models perform much better on coding problems released before their pre-training data cut-off.
Their words
The time when benchmarks lasted multiple decades has passed. Going forward, we will rely less on public benchmark results.
From one piece AI Drugs So Far 2 beliefs, in the piece's order there
korrents.com
A compound's AI origin story is not by itself a reason to expect it to do better in the clinic.Their words
For now, I am not convinced that issuing press releases about your compounds that talk about their discovery through AI techniques is sufficient to expect greater things from them.
korrents.com
The drugs advertised as having AI-discovered targets are aimed at targets people already knew about.Their words
What you will see is that in almost every case, these targets were already known to be implicated in the disease under investigation.
Their words
Games are AI’s natural habitat.
From one piece The case for ensuring that powerful AIs are controlled 5 beliefs, in the piece's order there
Their words
That is, labs should make sure that the safety measures they apply to their powerful models prevent unacceptably bad outcomes, even if the AIs are misaligned and intentionally try to subvert those safety measures.
Their words
The basic problem with evaluating alignment is that no matter what behaviors you observe, you have to worry that your model is just acting that way in order to make you think that it is aligned.
Their words
So when evaluating control, we should count catching an AI red-handed as a win condition.
Their words
AI control (with only black-box techniques) seems like a fundamentally limited approach.
Their words
We're advocating that companies handle risk from scheming models in a similar way–striving to ensure that they'll be safe even if their alignment efforts fail to prevent models from scheming.
From one piece Sean Carroll: General Relativity, Quantum Mechanics, Black Holes & Aliens | Lex Fridman Podcast #428 3 beliefs, in the piece's order there
Their words
What a large language model is trying to do is to predict text. That’s what it does. And it is leveraging the fact that we human beings for very good evolutionary biology reasons, attribute intentionality and intelligence and agency to things that act like human beings.
Their words
And rather than trying to ask, how close is it to being a human-like intelligent, we should appreciate it for what its capabilities are, and that will both be more accurate and help us put it to work and protect us from the dangers better rather than always anthropomorphizing it.
Their words
I think this is one of the biggest things that physics can help with, and it’s an obvious kind of low-hanging fruit situation where the heat generation, the inefficiency, the waste of existing high-level computers is nowhere near the efficiency of our brains. It’s hilariously worse, and we haven’t tried to optimize that hard on that frontier.
From one piece How to Make Off Grid Data Centers Affordable 5 beliefs, in the piece's order there
Their words
The cooling systems in most data centers are energy hogs and contribute no direct value.
How to Make Off Grid Data Centers Affordableaustinvernon.site
Their words
Efforts to improve efficiency and remove vestigial features pay off handsomely.
How to Make Off Grid Data Centers Affordableaustinvernon.site
Their words
Deleting the UPSs could knock off ~$750/kW after considering equipment, installation, and square footage reduction.
How to Make Off Grid Data Centers Affordableaustinvernon.site
Their words
The 100-megawatt facility might have ten 10-megawatt blocks served directly by a 1500V DC feeder line without connections to other blocks that would require extra switches and routing.
How to Make Off Grid Data Centers Affordableaustinvernon.site
Their words
Some efforts, like nuclear-powered data centers, are reminiscent of Toyota's hydrogen push.
How to Make Off Grid Data Centers Affordableaustinvernon.site
korrents.com
Retrieval-augmented generation is the most practical approach for reducing LLM hallucinations.Their words
Retrieval-augmented generation (RAG; Lewis et al., 2020), which conditions on the LLM's generation on retrieved documents is the most practical paradigm IMO.
From one piece The Data Center is the New Compute Unit: Nvidia's Vision for System-Level Scaling 2 beliefs, in the piece's order there
Their words
The copper backplane in the data center rack is effectively the new advanced packaging in the system-level Moore’s Law race.
korrents.com
Air cooling has run out of headroom for AI racks, so liquid cooling is now a requirement rather than an upgrade.Their words
The critical change that makes this all possible is liquid cooling.
korrents.com
Which products become visible is decided by distribution friction rather than by any contest of merit.Their words
Your product might be great, but how do you get it on the shelves of Walmart? Or how do you get it on the shelves of Google? Because if you're not on the shelves of Walmart, you're not on the shelves of Google, you might be invisible. And it's friction that kept you there, not some sort of merit contest.
Their words
I continue to not want super-voting control over OpenAI. I never have. Never have had it, never wanted it. Even after all this craziness, I still don’t want it. I continue to think that no company should be making these decisions, and that we really need governments to put rules of the road in place.
Their words
the embodied, interactive, and multi-modal nature of first language (L1) acquisition is challenging to replicate with current models.
Their words
And oftentimes it's more important to them to have the public perception that they're good directors so they get the next best deal. If they have a reputation for taking on management too aggressively, word will get out in the small community of founders and they'll miss the next Google.
Their words
This lack of knowledge sharing may impede progress in AI development.
From one piece Encyclopaedia Automatica 2 beliefs, in the piece's order there
korrents.com
AI-generated writing is already fairly good but still lacks the direction and insight of human editors.Their words
As of today, computers write pretty good articles but lack the direction and insight of human editors.
korrents.com
AI should be approached as a collaboration between humans and computers, not a forced choice between them.Their words
We need to find ways to work with this new tech to get the best of humans and the best of computers, together, rather than thinking in terms of one or the other.
korrents.com
Expecting AI or longer lives to arrive in time makes people skip having a plan for falling birth rates at all.Their words
And they don’t see a need for a plan B, in case their forecast fails.
Recommendsrcmnd.app
SupervisedTheir words
Longtime tech reporter Matt Lynley offers an uncommonly technical assessment of new AI developments in his newsletter Supervised. Original reporting and a fast cadence help this one stand out from other AI newsletters.
Mixed onrcmnd.app
SuperintelligenceTheir words
the back half I think it goes off the rails and makes a ton of assumptions
Recommendsrcmnd.app
Neural Network Methods for Natural Language ProcessingTheir words
I also read a ton of technical books last year, but the most impactful one for me was Neural Network Methods for Natural Language processing by Yoav Goldberg
From one piece How Effective Altruism Lost Its Way 2 beliefs, in the piece's order there
Their words
Diverting attention and resources from global health and poverty is an enormous gamble, as it will make many lives poorer, sicker, and shorter in the name of fending off threats that may or may not materialize.
Their words
Unlike other interventions EA has sponsored, there are scant metrics for tracking the success or failure of investments in existential risk mitigation.
From one piece Jeff Bezos: Amazon and Blue Origin | Lex Fridman Podcast #405 3 beliefs, in the piece's order there
Their words
We do know that humans are doing something different from these models, in part because we're so power efficient. The human brain does remarkable things and it does it on about 20 watts of power. And the AI techniques we use today use many kilowatts of power to do equivalent tasks.
Their words
And so the only way to improve safety is to have an escape system. And historically, human-rated rockets have had escape systems. Only the space shuttle did not, but Apollo had one. All of the previous Gemini, etcetera, they all had escape systems.
Their words
It's interesting to me that large language models in their current form are not inventions, they're discoveries. The telescope was an invention, but looking through it at Jupiter, knowing that it had moons, was a discovery.
Their words, before
precisely because smart people do devote brain-cycles to these possibilities, the rest of us have correspondingly less need to.
Their words now
I accepted what’s turned into a two-year position at OpenAI, thinking about what theoretical computer science can do for AI safety.
Their words, before
If I were forced at gunpoint to guess, I'd say that human-level AI seemed to me like a slog of many more centuries or millennia
Their words now
I was wrong, of course, not to contemplate more seriously the prospect that AI might enter a civilization-altering trajectory, not merely eventually but within the next decade.
Their words, before
If I were forced at gunpoint to guess, I’d say that human-level AI seemed to me like a slog of many more centuries or millennia (with the obvious potential for black swans along the way).
Their words now
I was wrong, of course, not to contemplate more seriously the prospect that AI might enter a civilization-altering trajectory, not merely eventually but within the next decade.
Their words
Large language models are more or less generalizations of stuff that we have. There’s still breakthroughs in AI waiting to happen, and maybe they are happening and maybe they’ll be good, maybe not, but that’s not quite the same.
From one piece In Continued Defense Of Effective Altruism 2 beliefs, in the piece's order there
Their words
And I notice that the tiny handful of people capable of caring about 200,000 people dying of neglected tropical diseases are the same tiny handful of people capable of caring about the next pandemic, or superintelligence, or human extinction.
In Continued Defense Of Effective Altruismastralcodexten.com
Their words
EA has done a great job working on this (see list of accomplishments above), and I think the AI and x-risk people have just as much to be proud of as the global health and animal welfare people.
In Continued Defense Of Effective Altruismastralcodexten.com
From one piece Elon Musk: War, AI, Aliens, Politics, Physics, Video Games, and Humanity | Lex Fridman Podcast #400 2 beliefs, in the piece's order there
Their words
I think it would be wise for us to have at least an objective third party who can be like a referee that can go in and understand what the various leading players are doing with AI, and even if there’s no enforcement ability, they can at least voice concerns publicly.
Their words
If an AI cannot figure out new physics, it’s clearly not equal to humans, nor has surpassed humans because humans have figured out new physics.
From one piece Managing extreme AI risks amid rapid progress (with 24 co-authors) 4 beliefs, in the piece's order there
Their words
Despite evaluations, we cannot consider coming powerful frontier AI systems "safe unless proven unsafe". With current testing methodologies, issues can easily be missed. Additionally, it is unclear if governments can quickly build the immense expertise needed for reliable technical evaluations of AI capabilities and societal-scale risks. Given this, developers of frontier AI should carry the burden of proof to demonstrate that their plans keep risks within acceptable limits.
Their words
Without sufficient caution, we may irreversibly lose control of autonomous AI systems, rendering human intervention ineffective. Large-scale cybercrime, social manipulation, and other harms could escalate rapidly. This unchecked AI advancement could culminate in a large-scale loss of life and the biosphere, and the marginalization or extinction of humanity.
Their words
For AI to be a boon, we must reorient; pushing AI capabilities alone is not enough. We are already behind schedule for this reorientation. The scale of the risks means that we need to be proactive, as the costs of being unprepared far outweigh those of premature preparation.
Their words
Society's response, despite promising first steps, is incommensurate with the possibility of rapid, transformative progress that is expected by many experts. AI safety research is lagging. Present governance initiatives lack the mechanisms and institutions to prevent misuse and recklessness, and barely address autonomous systems.
From one piece Mark Zuckerberg: First Interview in the Metaverse | Lex Fridman Podcast #398 3 beliefs, in the piece's order there
korrents.com
Building an AI that embodies a real person is a much harder problem than inventing a new AI character from scratch.Their words
So that's a lot of what I think we need to nail for the creators, and that's why that one is actually a much harder problem, I think, than starting with new characters that you're creating from scratch.
Their words
So at some point, I think making it so that you could build an AI version of yourself that could interact with people not after you die, but while you're here to help people fulfill this desire to interact with you and your desire to build a community.
Their words
I don't think we necessarily want there to be one big super intelligence. We want to empower everyone to both have more fun, accomplish their business goals, just everything that they're trying to do. We don't tend to have one person that we work with on everything, and I don't think in the future we're going to have one AI that we work with.
Recommendsrcmnd.app
ChatGPT Custom InstructionsTheir words
Using Custom Instructions properly can completely transform your experience with ChatGPT.
From one piece Self-exfiltration is a key dangerous capability 2 beliefs, in the piece's order there
Their words
If a model was capable of self-exfiltration, it would have the option to remove itself from your control.
Self-exfiltration is a key dangerous capabilityaligned.substack.com
Their words
Moreover, model exfiltration is likely impossible to reverse.
Self-exfiltration is a key dangerous capabilityaligned.substack.com
Their words
What we don't want to happen with AGI, we want to happen with synthetic biology. What we don't want to happen online and software with language, we want for it to happen with bio-based materials.
From one piece Language models surprised us 3 beliefs, in the piece's order there
Their words
Most importantly, ML experts and superforecasters both seem to be massively underestimating future spending on training runs.
Their words
Massively underestimating near-future progress could be very risky.
korrents.com
AI progress is faster than people expect, and ordinary scaling can be enough to solve problems that looked very hard.Their words
This rapid scaleup will probably drive another qualitative leap forward in capability like what we saw over the last 18 months.
Their words, before
Reading the book had not changed my fundamental belief that, in balance, we were on a good trajectory with AI research: good for science and good for society with expected positive impacts in many domains.
Their words now
I now believe that I was wrong and short-sighted to ignore that dual-use nature. I also think I was not paying enough attention to the possibility of losing control to superhuman AIs.
Their words
Instead of decades to centuries, I now see it as 5 to 20 years with 90% confidence.
Their words
ChatGPT is fast-tracking the commodification of the human spirit by mechanising the imagination. It renders our participation in the act of creation as valueless and unnecessary.
From one piece Microincentives and Enshittification 2 beliefs, in the piece's order there
Their words
The only way for Google Search to grow is to make itself worse.
korrents.com
Ranking first in a search has become near enough to being true, which is why search and spam are permanently at war.Their words
Google’s the first — and often the last — place we go to for answers, so snagging top billing in a Google search results page is, approximately speaking, equivalent to being “true.”
From one piece George Hotz: Tiny Corp, Twitter, AI Safety, Self-Driving, GPT, AGI & God | Lex Fridman Podcast #387 5 beliefs, in the piece's order there
Their words
Of course, when you talk to an AI that’s made by a big company in the cloud, the AI fundamentally is aligned to them, not to you. And that’s why you have to buy a tiny box. So you make sure the AI stays aligned to you.
Their words
But the real thing that I want is not something that like tab completes my code and gives me ideas. The real thing that I want is a very intelligent pair programmer that comes up with a little popup saying, “Hey, you wrote a bug on line 14 and here’s what it is.”
Their words
I think things like MuZero and AlphaGo are so much more impressive because these things are playing beyond the highest human level. The language models are writing middle school level essays and people are like, wow, it’s a great essay.
Their words
I think future LLMs are going to be smaller, but are going to run looping on themselves and are going to have retrieval systems. And the thing about using a retrieval system is you can cite sources, explicitly.
Their words
What’s ironic about all these AI safety people is they’re going to build the exact thing they fear. We need to have one model that we control and align. This is the only way you end up paper clipped. There’s no way you end up paper clipped if everybody has an AI.
Nothing matches.
What is a korrent?
A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.
Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.
Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.
About the English under a post
Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.
The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.
Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.