Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI 2040: Plan A

ai-2040.com

Post reply on HN
Print view View original

561–567 of 567 posts

Re: AI 2040: Plan A

#561

Post by YeGoblynQueenne »

red75prime wrote: post #559
YeGoblynQueenne wrote: post #557

Earlier quoted context omitted.

Unfortunately those don't really work: https://youtu.be/C4NQNeSO2vs?si=epkxhVXpypOCppGW Also, none of those companies' cars are really autonomous. Waymo, for example, relies on remote workers that are ready to intervene and suggest a course of action when the AI driver gets stuck: https://waymo.com/blog/2024/05/fleet-response/ And the point is that the AI driver gets stuck because it can't understand the situation it…

> https://youtu.be/C4NQNeSO2vs?si=epkxhVXpypOCppGW "VLAs fine-tuned on human demonstrations overfit. Here's how to mitigate it." Er, OK, I guess. But I think that Waymo uses mitigations or a different approach (RL, for example).

Unfortunately the mitigations in the video don't work either.

The bottom line is that none of the techniques we have have been shown to work, to "really" really work, in the real world. If they did, you'd know instantly because suddenly robot cars and robot maids would be absolutely everywhere, not just in a few cities in the States or China. The Chinese in particular are chomping at the bit to flood the market with cheap robot labour. They can't because there's no such thing, yet. Not because there aren't cheap robots: because there's not the AI to drive them.

I know it's hard to get a good feel for where the state of the art is currently, but try this exercise. Back in 2017, if you were watching the video of Atlas doing backflips [1], how far would you think we are from robotic autonomy? Most people would have said "very close" and indeed many did. Yet, 9 years later we still don't have robot maids and robot butlers.

Atlas, of course, was hand-coded, but that doesn't make any difference. Its programming was good enough for backflips (and many other really impressive feats besides; e.g. much more impressive was seeing it balance itself after stepping on a length of pipe, but I can't find the video now). Just because there's a trend now to do the same coding by machine learning rather than by hand, doesn't change anything. The point is we don't know how to create the right program for a robot to act autonomously, independently, freely, and safely, in the real world.

Autonomy is hard and it goes way beyond anything anyone has demonstrated in the last... I'd say 50 years. Robotics has a long history of very impressive results that never turn out as useful as people imagine.

_________________

[1] https://www.youtube.com/watch?v=DqdaT9LI5oE

Re: AI 2040: Plan A

#562

Post by red75prime »

YeGoblynQueenne wrote: post #560
red75prime wrote: post #558

Earlier quoted context omitted.

Non-trivial percentage of people work in the real world until they do something stupid and work no more. It's not real autonomy, not yet. It's a matter of degree. Sorry, I don't want to watch an hour long video to be told how something that works 99.99% of the time doesn't really work for some contrived definition of "really". BTW, as is typical with people, remote operators occasionally cause problems.

Fair if you don't want to watch the entire video. The VLA stuff is early in the video but I can summarise it for you (if you trust me to do so): basically none of the current deep learning based approaches to autonomy generalise. Not RL, not transformers, not anything else. They all tend to work fine in environments and tasks within their training set but outside of it, pffft, performance evaporates. That's my summar…

> So what if your image classifier thinks a cat is an impala?

For starters, the artificial neural network is the only instrument we have for sufficiently robust image classification. If it doesn't work nothing else will. We can only validate and retrain.

Getting back to self-driving. What if an occupancy network misclassify an impassable surface patch as passable? It would mean a possible accident, so you bias the network towards impassable classification. In this way misclassification is overwhelmingly more likely to cause spurious braking, unneeded swerve, and the like rather that a collision ("brake stabbing" was a recurring problem in Tesla FSD. Today it is largely mitigated.)

> The fact they can't is the strongest evidence that whatever proprietary secret sauce moat they have... doesn't really work.

Well, Waymo has publicly embraced world models, VLMs and end-to-end backpropagation relatively recently (December 9, 2025)[1]. There's not enough data yet to judge how this has influenced their expansion rate. Waymo representatives tout improved generalization (sorry, I can't find the interview. I think it was Dmitri Dolgov.)

> if their systems were really autonomous they could just deploy them anywhere a human driver can drive

As I've said it's a matter of degree. Human-caused road casualties are a familiar and largely inevitable evil. An autonomous vehicle killing a person will (and already has) cause a media storm. They can't ignore a potentially problematic distribution shift caused by deployment in a new area.

You are talking about the perfect system. Real systems are computationally constrained. Distillation necessarily degrades performance and might require retraining for particular areas.

[1] https://waymo.com/blog/2025/12/demonstrably-safe-ai-for-auto...

Re: AI 2040: Plan A

#563

Post by YeGoblynQueenne »

red75prime wrote: post #562
YeGoblynQueenne wrote: post #560

Earlier quoted context omitted.

Fair if you don't want to watch the entire video. The VLA stuff is early in the video but I can summarise it for you (if you trust me to do so): basically none of the current deep learning based approaches to autonomy generalise. Not RL, not transformers, not anything else. They all tend to work fine in environments and tasks within their training set but outside of it, pffft, performance evaporates. That's my summar…

> So what if your image classifier thinks a cat is an impala? For starters, the artificial neural network is the only instrument we have for sufficiently robust image classification. If it doesn't work nothing else will. We can only validate and retrain. Getting back to self-driving. What if an occupancy network misclassify an impassable surface patch as passable? It would mean a possible accident, so you bias the ne…

I largely agree with what you say, except the bit where you say I'm talking about the perfect system. I tried hard not to give that impression because that's not what I'm on about. Autonomous systems trained with deep learning (including Deep RL from human demonstrations and so on) are very far from "perfect". They can perform very well in a broad range of scenarios until they encounter an unforeseen situation and their performance falls off a cliff. The problem is that the real world is full of unforeseen situations, full of long-tailed distributions that are near impossible to learn just by training on data, or anyway a reasonable amount of data (even very big data).

Sure, you can "mitigate" whatever edge case you become aware of during operation but when the edge cases are the entire distribution there's always more of them where the ones you mitigated for came from and that's a big problem, it's a very real problem, and we don't yet know how to solve it.

>> Well, Waymo has publicly embraced world models, VLMs and end-to-end backpropagation relatively recently (December 9, 2025)[1]. There's not enough data yet to judge how this has influenced their expansion rate. Waymo representatives tout improved generalization (sorry, I can't find the interview. I think it was Dmitri Dolgov.)

Well, let's wait and see. But my intuition is that Waymo is simply trying the latest trend to see if it will change something. It might even move the needle a bit, who knows.

Re: AI 2040: Plan A

#564

Post by gamblor956 »

gabriel-uribe wrote: post #491
gamblor956 wrote: post #482

Earlier quoted context omitted.

Most companies have already given up on AI. It was a bigger disappointment than big data. The only companies sticking with AI at this point are the major players who have chasing their masssive overinvestments to the bitter end, crappy coding shops, and for some reason Starbucks and they'll all suffering badly in a year their customers begin demanding massive amounts of human coding to repair all the issues with the…

We're still in the early adoption phase. Betting against the internet wasn't a great idea, and betting against AI doesn't look like one either ;)

We're past the early adoption phase. We're in the same trough of disillusionment phase that big data went through.

Re: AI 2040: Plan A

#565

Post by groundzeros2015 »

mrdomino- wrote: post #110
ekelsen wrote: post #107

Earlier quoted context omitted.

How is this different from farming? It takes the limited resources of land and water from a community and sells the result for profit as food or fuel. The vast majority of profit is made downstream and outside the community. Golf courses being a traditional green place where people gather seems a bit far fetched to me when most of them are elite private clubs.

In farming, the result is food that can be eaten.

they are one provider among many. Everybody needs food. But nobody needs that particular farm's food.

Re: AI 2040: Plan A

#566

Post by layla5alive »

disgruntledphd2 wrote: post #541
layla5alive wrote: post #535

Earlier quoted context omitted.

They already possess general intelligence by many metrics. Sure they miss a few, but that's nitpicky and goalpost shifty - lots of humans make errors of all sorts as well, or incur brain damage limiting them in one or a few areas of intelligence - we don't then say they are not general intelligence anymore. I think maybe you mean superintelligence, which is a more fair critique.

> They already possess general intelligence by many metrics. Can you share the metrics you are using for this assessment? They are really powerful tools, but a quick glance at their thinking tokens (which is a bad name, tbh) rapidly disabuses me of the notion that they are general intelligences. They possess large amounts of crystallised intelligence (i.e. they have a lot of knowledge), but their fluid intelligence i…

My take is that fluid intelligence (quite a bit) below human median is still general intelligence (that is, general intelligence doesn't mean no gaps).

It feels like we have collectively goal-post shifted the definition of AGI to be closer to that of ASI.

One of those "what do you call a doctor who graduated at the bottom of their class" type things. I think despite their genuine deficits (and there are many), frontier LLMs have basically cleared the minimum AGI bar.

Though I realize a lot of people don't agree with this take :)

Re: AI 2040: Plan A

#567

Post by layla5alive »

esafak wrote: post #553
layla5alive wrote: post #533

Earlier quoted context omitted.

No, they're proposing a spying panopticon and state control of global resource distribution - specifically general purpose compute - including seizing and destroying GPUs. They're proposing a totalitarian global dictatorship controlling computing hardware and software. Lest you think I'm being hyperbolic: https://ai-2040.com/supplements/covert-ai-projects This is arsonists selling fire insurance.

A nasty global inspections regime just like we have for nuclear weapons, which are less dangerous than this. Oh no ...

Nuclear weapons are less dangerous than compute? Are you listening to yourself?

Humans employing this kind of thinking and desire to control other people is the genuine danger here.

Post reply on HN