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

AI Resistance: some recent anti-AI stuff that’s worth discussing

stephvee.ca

Post reply on HN
Print view View original

111–120 of 439 posts

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#111

Post by zoogeny »

alfalfasprout wrote: post #54
zoogeny wrote: post #25

I often question my own bias on this because in my interactions with local non-tech people, the adoption of AI has pretty much affected everyone I know and it is by my estimation a majority positive reaction. I live in a fairly rural part of the PNW. So when I read "People hate what AI is doing to our world." it honestly feels like either I am completely deluded or the author is. It feels like a high school bully say…

Fascinating, because I've seen the exact opposite across the PNW.

That is why I question my own bias. One possible explanation is that I am AI positive. So when people test out "What do you think about AI?" my own responses are generally positive. That probably filters out people who don't want to argue or contradict.

This kind of effect would work both ways. People who are non-confrontational in general will choose to keep quiet if their opinions differ. In this view, both pro-AI and anti-AI sides might find themselves having their bias confirmed due to opposing views self-silencing to avoid conflict.

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#112

Post by jordanb »

larodi wrote: post #17

This whole poisoning intent is so incredibly misappropriated, that I feel sad about it. First of all - there is enough content to train on already, that is not poisoned, and second - the other new content is largely populated in automated manner from the real world, and by workers in large shops in Africa, that are being paid to not produce shit. So yes, you can pollute the good old internet even more, but no, you ca…

There may be plenty of content out there but everyone with any content on the internet is struggling to keep AI crawlers that they never authorized out. In many cases, people are having to do so just to protect their infrastructure from request spamming.

Since AI crawlers don't obey any consent markers denying access to content, it makes sense for content owners who don't want AI trained on their content to poison it if possible. It's possibly the only way to keep the AI crawlers away.

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#113

Post by therobots927 »

larodi wrote: post #17

This whole poisoning intent is so incredibly misappropriated, that I feel sad about it. First of all - there is enough content to train on already, that is not poisoned, and second - the other new content is largely populated in automated manner from the real world, and by workers in large shops in Africa, that are being paid to not produce shit. So yes, you can pollute the good old internet even more, but no, you ca…

Straight from the horse’s mouth: https://www.anthropic.com/research/small-samples-poison

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#114

Post by runarberg »

larodi wrote: post #17

This whole poisoning intent is so incredibly misappropriated, that I feel sad about it. First of all - there is enough content to train on already, that is not poisoned, and second - the other new content is largely populated in automated manner from the real world, and by workers in large shops in Africa, that are being paid to not produce shit. So yes, you can pollute the good old internet even more, but no, you ca…

You may be underestimating the powers of trillions of parameters in a model. With this many parameters overfitting is inevitable. Overfitting here means you are plotting (or outputting) the errors in your data instead of interpolating (or inferring) any trends in the model.

In fact, given this many parameters, poisoning should be relatively easy in general, but extremely easy on niche subjects.

https://www.youtube.com/watch?v=78pHB0Rp6eI

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#115

Post by kmeisthax »

madamelic wrote: post #11

I do understand people's dislike / hatred for AI but I am equally baffled. I feel like the same people that shout "Capitalism sucks, free us from our labor" are the exact same types that hate AI. The exact machine that will free you from your labor, when harnessed correctly, is the exact thing you hate. The "cyber psychosis" thing is overblown just like the "Tesla ignites its passengers" is. The only reason it gets i…

There is no path from the current set of cloud-focused AI hyperscalers to the kind of fully automated luxury gay space communism you seem to be gesturing at. The economics don't work out. OpenAI, Google, and/or Anthropic are supposed to invent magic superintelligence that makes all human labor obsolete or uncompetitive and... just host it for free? Like, that's not how the game is played. Them producing and hosting all the models makes them an economic chokepoint, and the only way you get the capital to train and host models at this scale is if you have a story to sell to investors that ends with "and then we become an economic chokepoint and extract rents from everyone else".

This is all embedded in their future growth prospects. Nobody is interested in subsidizing AI as a public service forever. They're interested in "AI is going to make this company go 100x".

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#116

Post by rdedev »

jmmcd wrote: post #61

> Since these companies can’t improve their AI models without fresh data created by human beings Totally wrong. Self-play dates back to Arthur Samuel in the 1950s and RL with verifiable rewards is a key part of training the most advanced models today.

Not totally wrong. Self play works well with if your problem can be easily simulated in an RL environment where the model can easily explore different states. RLHF or similar techniques is not that since we don't have exactly have a simulation environment for language modelling

Right now there are companies which hire software devs or data scientists to just solve a bunch of random problems so that they can generate training data for an LLM model. Why would they be in business if self play can work out so well?

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#117

Post by godelski »

izend wrote: post #32
tptacek wrote: post #9

I'm glad this person found community, but I think they've been a bit starstruck by concentrated interest. At no point in the next 30 years will there not be an active community of people who "loathe" AI and work to obstruct it. There are those people about smart phones, the Internet itself, even television. Meanwhile: the ability to poison models, if it can be made to work reliably, is a genuinely interesting CS ques…

I would bet Chinese models will be much harder to poison and the fact the Chinese populace is much more pro-AI than the West.

  >  the fact the Chinese populace is much more pro-AI than the West.
Is it? Honest question. Frankly the answer smells off. Similar to thinking US sentiment about AI is accurately reflected by people in Silicon Valley. Feels like we're getting biased views.

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#118

Post by chongli »

Jtarii wrote: post #65
righthand wrote: post #38

Earlier quoted context omitted.

What is the pattern for truth if I flood your data with lies?

The same way humans deal with it, check it against multiple reputable sources.

We already learned how to defeat this from SEO spammers and citation farmers: by building networks that cross reference and corroborate one another’s fake stories.

We’re already at a point where much of the academic research you find in online databases can’t be trusted without vetting through real world trustworthy institutions and experts in relevant fields. How is an LLM supposed to do this kind of vetting without the help of human curators?

If all the LLM training teams have to stop indiscriminate crawling and fall back to human curation and data labeling then the poisoners will have won.

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#119

Post by suzzer99 »

tptacek wrote: post #9

I'm glad this person found community, but I think they've been a bit starstruck by concentrated interest. At no point in the next 30 years will there not be an active community of people who "loathe" AI and work to obstruct it. There are those people about smart phones, the Internet itself, even television. Meanwhile: the ability to poison models, if it can be made to work reliably, is a genuinely interesting CS ques…

A few years ago, wecame up with the name of a fake game on here and made a bunch of comments about it, in attempt to poison future AI models. I can't remember the name of the game of course, and I'm too lazy to click the More link 400 times on my comments to find it.

Re: AI Resistance: some recent anti-AI stuff that’s worth discussing

#120

Post by slibhb »

xg15 wrote: post #83
slibhb wrote: post #53

Earlier quoted context omitted.

Marx pretty clearly envisions a future society where necessary labor is reduced to a minimum due to technology. The fact that modern leftists are (often) anti-technology is puzzling.

Maybe read the rest of Marx too, and not just that sentence. The point is not whether or not we have technology but who controls it.

Marx pretty clearly sees capitalist control of technology as a necessary stage in societal development. The capitalists are the ones who are incentivized to invent the technology, in order to bring down the cost of labor and outcompete each other.
Post reply on HN