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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
Retrieval-augmented generation (RAG; Lewis et al., 2020), which conditions on the LLM's generation on retrieved documents is the most practical paradigm IMO.
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.
So part of the reason that we deploy the way we do, we call it iterative deployment, rather than go build in secret until we got all the way to GPT-5, we decided to talk about GPT-1, 2, 3, and 4. And part of the reason there is I think AI and surprise don’t go together. And also the world, people, institutions, whatever you want to call it, need time to adapt and think about these things.
I would heavily discourage any startup that was thinking about starting as a nonprofit and adding a for-profit arm later. I’d heavily discourage them from doing that. I don’t think we’ll set a precedent here.
That’s a theatrical risk. That is a thing that can really take over how people think about this problem. And there’s a big group of very smart, I think very well-meaning AI safety researchers that got super-hung up on this one problem, I’d argue without much progress, but super-hung up on this one problem. I’m actually happy that they do that, because I think we do need to think about this more. But I think it pushed out of the space of discourse a lot of the other very significant AI- related risks.
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.
Several people have speculated that we'll see many very valuable one-person companies (see Sam Altman's interview and Reddit discussion). I think they might be right.
One pattern is that for simple prompts, weak models can do (nearly) as well as strong models. For more challenging prompts, however, users are much more likely to prefer stronger models.
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.
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.
Lots of ML techniques in the post can help with data quality, but fundamentally human data collection involves attention to details and careful execution.
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.
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
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.
Unlike other interventions EA has sponsored, there are scant metrics for tracking the success or failure of investments in existential risk mitigation.
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.
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.
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.
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.
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.
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.
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.
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.
I tried out Humane's long-awaited new AI pin. While there are some undoubtedly novel new features in its interfaces, I think "AI hardware" may turn out to be a dead end
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.”
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.
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.”
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.
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.
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.
Thus, the 2nd UI paradigm will survive, albeit in a less dominant role. Future AI systems will likely have a hybrid user interface that combines elements of both intent-based and command-based interfaces while still retaining many GUI elements.
The third UI paradigm, represented by current generative AI, is intent-based outcome specification. The user tells the computer the desired result but does not specify how this outcome should be accomplished, such as the steps to be executed. Compared to traditional command-based interaction design, this completely reverses the locus of control.
I doubt that the current set of generative AI tools, like ChatGPT, Bard, etc., are representative of the UIs we’ll be using in a few years because they have deep-rooted usability problems.
A funny book explaining the basics of AI! The subtitle is How Artificial Intelligence Works and Why It's Making the World a Weirder Place. A great introduction to AI. With a cute cartoon mascot. The title is from her training an AI to write romantic greeting cards.
My response is that their position is non-scientific – What is the testable hypothesis? What would falsify the hypothesis? How do we know when we are getting into a danger zone?
I even think AI is going to improve warfare, when it has to happen, by reducing wartime death rates dramatically. Every war is characterized by terrible decisions made under intense pressure and with sharply limited information by very limited human leaders.
My view is that the idea that AI will decide to literally kill humanity is a profound category error. AI is not a living being that has been primed by billions of years of evolution to participate in the battle for the survival of the fittest, as animals are, and as we are. It is math – code – computers, built by people, owned by people, used by people, controlled by people.
But we do not have enough data and we do not have enough insight to use AI/ML to pick better targets that have a higher chance of succeeding in the clinic.
The app costs a few dollars up front and then has an optional subscription for unlimited requests, but I would recommend going directly to OpenAI’s developer dashboard, creating your own account, and copy/pasting your own API key into Petey.
This isn't perfect, but soon it will be, and you will be able to listen to or read any content you want in the timeframe, tone, voice and bias of your choosing.
Hopefully I've convinced you that chatbots are a terrible interface for LLMs. Or, at the very least, that we can add controls, information, and affordances to our chatbot interfaces to make them more usable.
I want to see more tools and fewer operated machines - we should be embracing our humanity instead of blindly improving efficiency. And that involves using our new AI technology in more deft ways than generating more content for humans to evaluate. I believe the real game changers are going to have very little to do with plain content generation.
Good tools let the user choose when to switch between implementation and evaluation. When I work with a chatbot, I'm forced to frequently switch between the two modes.
The biggest issue is probably that we don’t control neutral networks enough to be able to ensure AI doesn’t harm humans. We can’t even control AI to not reveal internal prompts.
This book is my best crack at explaining what I think is an existential risk to liberal societies and what I think we need to do to get to that awesome future I used to be so excited about.
What ChatGPT is, in this instance, is replication as travesty. ChatGPT may be able to write a speech or an essay or a sermon or an obituary but it cannot create a genuine song.
I think this matches the general finding that AI progress is faster than expected, and increases my certainty that scale and normal progress can sometimes be enough to solve even very difficult problems.
But in terms of what we value the most as humans, which is to say our feelings, our emotions, our sense of what the world is in a very personal way that I think means as much or more to people than their information processing. And that’s where I don’t think that AI necessarily will become conscious because I think it’s the property of life.
The best and easiest-found-by-optimization algorithms for solving problems we want an AI to solve, readily generalize to problems we’d rather the AI not solve; you can’t build a system that only has the capability to drive red cars and not blue cars, because all red-car-driving algorithms generalize to the capability to drive blue cars.
For example, large language models can generate outputs that are untruthful, toxic, or simply not helpful to the user. In other words, these models are not aligned with their users.
This is because the language modeling objective used for many recent large LMs-predicting the next token on a webpage from the internet-is different from the objective "follow the user's instructions helpfully and safely" (Radford et al.,, 2019; Brown et al.,, 2020; Fedus et al.,, 2021; Rae et al.,, 2021; Thoppilan et al.,, 2022). Thus, we say that the language modeling objective is misaligned.
Even though InstructGPT still makes simple mistakes, our results show that fine-tuning with human feedback is a promising direction for aligning language models with human intent.
Science journalist Jacob Ward warns about a future in which AI technology forms a tight cybernetic feedback loop with the biases ingrained into the human brain, pushing our day-to-day existence in frightening directions. It's sort of like a version of The Terminator in which Danny Kahneman plays a starring role. As longtime readers and listeners of mine know, a big theme in my techno-criticism is the under-appreciated degree to which technologies exert powerful, unintended consequences on our personhood and culture, so Ward's warning hits a sweet spot with me.
I I would I would not say that about Google. I would just say that more generally like that is the incumbent temptation. Um it is is basically is basically go from attack mode to defend mode, right? Go from being the pirate to being the navy.
I wouldn’t say that this is the most compelling book I’ve ever read in terms of the prose style or storytelling, but it does provide a very helpful, almost quantitative overview of all the potential threats looming out there
AI was in the air in the mid 1980s, but there were two things especially that made me want to work on it: a novel by Heinlein called The Moon is a Harsh Mistress
If you're curious how Google built it's culture and the processes thanks to their extremely data & engineering driven nature, this is a great book. What held it back was being ~100 pages too long and Bock didn't always understand where what they did only fit at a $100Bn+, mega-profitable company. Still, the insights on the studies on their 50,000+ employees is well worth the read since so few of us can get similar statistical significance to apply to our teams.
The interests of a Venture Capitalist are different than those of the entrepreneurs building a company they've invested in. Jeff does an awesome job of helping explain how you can get misaligned in your goals versus your investors. Fortunately, he also covers how to avoid it.
this channel is full of high production quality and well-written videos. There are also several series on machine learning, the topics underlying self-driving cars, so this is certainly one of the more applied math channels out there.
Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches.
This book stretched my mind more than almost any other book I’ve read. It’s tough at parts, it’s long, but you’ll come out of it thinking about brains, minds, intelligence, and AI in an entirely new way.
As generative drug analoging grows in importance, it's going to be crucial for people to make the entire training set available in detail when such work is published.
spending an increasing amount of my free time working with tools like Google Colab, Runway ML, TensorFlow, and Hugging Face to play around with machine-generated text, audio, imagery, and video.
We argue that ARC can be used to measure a human-like form of general fluid intelligence and that it enables fair general intelligence comparisons between AI systems and humans.
When I'm writing articles etc that need to be shared with an editor, I typically write in Microsoft Word. I hate all word processing software, and anyone who insists that Google Docs is so much better than Word is a liar.
It's increasingly coming to the fore and Marcus deals with it in a really interesting way. He takes us through lots of different areas where AI is being used to mimic creativity, which some people would argue is the essence of being human. It's interesting to see how close to being human AI is getting.
so we use Google Meet speakermics, which allow for full-duplex conversations, which makes conversation far more organic. That device turned Google Meet from something exhausting to something really good.
In terms of iOS apps, I use PDF Expert, which is a great PDF reader and annotator (using Apple Pencil), Notability as a virtual whiteboard (again with Apple Pencil), Google Docs for document editing, Google Calendar, and Gmail.
Eric Schmidt, Jonathan Rosenberg and Alan Eagle, Trillion Dollar Coach: The Leadership Playbook of Silicon Valley’s Bill Campbell: I understood Bill Campbell was a behind-the-scenes guy in Silicon Valley, but I had no idea just how influential he was. Bill Gurley of Benchmark noted “I would argue that Bill has had a bigger impact on Silicon Valley than any other single person simply because his reach was so amazingly wide.” That’s a story I want to read.
For coding, I use Visual Studio Code. I use Git for source control, Photoshop for 2D art, Audition for audio editing, Premiere for video editing, and our team coordinates on Google Docs and Skype.
What we are actually listening to is human limitation and the audacity to transcend it. Artificial Intelligence, for all its unlimited potential, simply doesn’t have this capacity.
Now, almost everything happens in Google Docs, except for when I write straight into the Axios or Substack CMS. (Both of which are lovely, WYSIWYG, very simple and easy to use.)
Such restrictions are sub-optimal for sentence-level tasks, and could be very harmful when applying fine-tuning based approaches to token-level tasks such as question answering, where it is crucial to incorporate context from both directions.
We argue that current techniques restrict the power of the pre-trained representations, especially for the fine-tuning approaches. The major limitation is that standard language models are unidirectional, and this limits the choice of architectures that can be used during pre-training.
We argue that current techniques restrict the power of the pre-trained representations, especially for the fine-tuning approaches. The major limitation is that standard language models are unidirectional, and this limits the choice of architectures that can be used during pre-training.
Such restrictions are sub-optimal for sentence-level tasks, and could be very harmful when applying fine-tuning based approaches to token-level tasks such as question answering, where it is crucial to incorporate context from both directions.
Such restrictions are sub-optimal for sentence-level tasks, and could be very harmful when applying fine-tuning based approaches to token-level tasks such as question answering, where it is crucial to incorporate context from both directions.
We argue that current techniques restrict the power of the pre-trained representations, especially for the fine-tuning approaches. The major limitation is that standard language models are unidirectional, and this limits the choice of architectures that can be used during pre-training.
We argue that current techniques restrict the power of the pre-trained representations, especially for the fine-tuning approaches. The major limitation is that standard language models are unidirectional, and this limits the choice of architectures that can be used during pre-training.
Such restrictions are sub-optimal for sentence-level tasks, and could be very harmful when applying fine-tuning based approaches to token-level tasks such as question answering, where it is crucial to incorporate context from both directions.
To me, this book is an illustration of the power of names. Today, in the era of Google, if you know the name of something, you can find out about it with a simple search. But if you don't know of what you're looking for, it suddenly becomes much harder to find it. Having in the back of your head the names of common algorithms that help you solve problems is really powerful.
Google Chrome: What's a Web Developer without Google Chrome browser? The first thing I do with Safari is to get Google Chrome and install the latest version.
Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy.
Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy.
In particular, this means that incredibly powerful AI will emerge in a world where crazy stuff is already happening (and probably everyone is already freaking out).
The main disagreement is not about what will happen once we have a superintelligent AI, it’s about what will happen before we have a superintelligent AI.
One of the most commonly emphasized points in the AI safety literature is that one must never anthropomorphize AI—that is, don’t project human mental properties onto artificially intelligent systems.
A superintelligence whose goal system is even slightly misaligned with ours could, being far more powerful than any human or human institution, bring about human extinction for the very same reason that construction workers routinely slaughter large populations of ants.
For writing code, I use emacs and Google's internal distributed build system (a version of which was open sourced as Bazel) and our version control system
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.