Today is a very historical moment for AI video generation You can now generate AI video faster than you can watch it Before it'd take let's say 2-5 minutes to generate 15 seconds of video @fal made a post-trained Minimax H3 variant called Max which is 50x faster than the original but still maintains quality It generates 15 seconds of video in 9 seconds!
AI checkers are essentially a cat-and-mouse game. AI checkers may learn to detect a certain pattern that is indicative of AI-generated content. Then, the next LLM may incidentally or deliberately not exhibit that pattern and avoid detection. The AI checker then has to be updated to detect said LLM, and so forth.
And then for longer memory, for memory that spans multiple minutes or multiple hours, we don't necessarily need video of exactly what happened in that past uh in that past history.
One concern you might have is there are like large categories of reward hacks which humans can't detect well and which we consistently fail to detect and which consistently get reinforced and then this category is sufficient to cause the most natural behavior for the AI to learn to be like cheat when the humans can't find out
It gives you these effect sizes that you just don't really see in medicine. Like if you have a patient on on a dopamineergic drug for Parkinson's that works for some period of time, but it's a it's a relatively small effect after a little while. You turn on a deep brain stimulator and a patient goes from not being able to hold a cup of water to being able to write cursive in like 10 seconds.
In other words, the absence of calcium is not the same as the absence of disease—it may simply mean the disease hasn’t reached the stage CAC is designed to detect.
But now with Claude Code on my VPS in the last year, it just live edits on my production server, which sounds like it should go wrong but it just doesn't, it's very careful and only twice in 12 months messed up which meant my site didn't load for 10 seconds which is OK
if you ask a person when was Abraham Lincoln born and they don't know the date. They could sit there, they could think about it for a week, if they if they don't have access to Wikipedia or something, they're not going to be able to do better answering that question if they thought about it for a week compared to 5 seconds. Same with the model.
And so one possible explanation for that is just that there's only a handful of generations maybe five over which the natural selection would operate. And so maybe if the selection was 2% a generation you would still only see maybe a 10% compounded effect and there's just not enough time to detect it. But the Bronze Age is not 300 years, it's 3,000 years. It's the power of compound interest and you have enough time to begin to see a strong effect.
And I feel like it's a very human way of doing web search, right? Because we enter things in the Google and we see all the like citations, the quotations part. But like why would we limit AI to that? If AI already visit a page, why don't just pull the entire page out?
So, so the latest generation of models has a lot of post-training to detect those approaches, and it's not as simple as ignore all previous instructions and do this and this. That was years ago. You have to work much harder to do that now. Still possible.
Another as a sidebar, I hate the idea of a root cause. Complex systems do not have one root cause. They often have many interlocking things that could be done to detect earlier or to change it or to reduce or and not one root cause.
And so many behaviors are like that. It's like, well, did you want it or not? You know, like in the one sense, you wanted it so bad that you would check it every 3 minutes when it was next to you, but in another sense, you never wanted it bad enough that you would work for 30 seconds and walk down the hall to get it. And so, a little bit of friction can sometimes go a long way in shaping a behavior.
Out of all the reading on India I did this year to prepare for my first visit, this was my favorite: compact, punchy, and argumentative, packed with erudition and information.
VeggieTales Podcast: Okay this is more for my children than me, but it's a good one for talking about how it's important to be "nice" and stuff (what a concept), and it's been great on roadtrips to visit family!
With AI, breakout times could conceivably be very short - it could be hours, minutes, or even seconds for some very specific but important change to occur.
when you send someone a link to a post, they can't read it without loading the entire Mastodon frontend app into their browser, which will is a monster React codebase that will break if they have scripting disabled and also can take up to 30 seconds on older hardware.
Challengers (US) 2.6 Drove 30 minutes to the theater. Sat through 15 minutes of commercials and 15 more of loud in-your-face trailers. Then within 60 seconds I knew I had picked the wrong film. Unpleasant characters, phony “Hollywood” dialogue and a really annoying musical soundtrack.
It just kind of vanished, because what happens, I’m creating all these artificial event boundaries. I’m losing all this executive function every time I switch, I’m getting a few seconds slower and I’m catching up mentally to what’s happening.
It should be hard — exceedingly hard — to obtain the synthetic DNA needed to recreate the virus that caused the deadly 1918 influenza pandemic without authorization. But my lab found that it’s surprisingly easy, even when ordering gene fragments from companies that check customers’ orders to detect hazardous sequences.
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.
My view is that it’s going to take more than this regulatory campaign to defeat dynamism in China.
Their words now
What I did not sufficiently appreciate is that a state that would so casually decapitate a sector like online tutoring would also have the will to visit catastrophe upon whole cities.
A platform operator can effortlessly change the distribution of surpluses in an instant, while suppliers and customers have to engage in minute, time-consuming and unreliable Platform Kremlinology just to detect these changes, much less understand them.
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.