"For helpful comments on drafts, and/or other assistance, I thank: ..." Odd to have Acknowledgements but no information at all about the author. Who are they?
"Copyright © 2023-2023 David Chapman" from the bottom of the page
Only You Can Stop an AI Apocalypse
31–40 of 130 posts
Re: Only You Can Stop an AI Apocalypse
#32"Most AI researchers think good outcomes are more likely. This seems just blind faith, though" Or, you know, they might actually know something about the subject? On a similar note, most physicists don't think particle colliders will cause mini black holes which will swallow the Earth, but the fear gets repeated by non-physicists.
You are fallaciously appealing to (assumed) authority and missing entirely the fact that the author is a PhD in AI. This critically underscores why blind faith in authority is bad, and why people should do some basic research such as reading the article. By reading the article you would have understood the arguments the author is making on their merits, rather than on blind faith in imagined authority figures you are citing.
The author is not talking about sentient AI turning Hollywood film evil like Elon Musk did. Why no, that would be ludicrous. You would have easily noticed that if your mental model didn’t get trained to filter out article titles like these…
In fact what you did with your mental model is exactly some of the problems with AI that is dangerous. There are more banal problems with AI such as encoding discrimination and automating it, possibly absolving people of the responsibility of discrimination and prejudice. AI in its current form is a lot of automating assumption making. Imagine taking your assumption making powers, and extracting it into an unaccountable model that can be deployed at national or global scales.
Re: Only You Can Stop an AI Apocalypse
#33Anti-AI seems poised to be the anti-vax of the next few years, where laymen pick up lines like "we have no idea how AI does the stuff it does" and take that wholly out of context. As a society, we have to decide how we feel about AI in legal and cultural contexts as we expand our capabilities, but there's time to do that. Github Copilot will not end up launching nuclear missiles in the meantime. Stuff like this just…
I had the same impression. He says things which are demonstrably untrue, it implies a lack of domain knowledge. But, on paper anyway, he has the domain knowledge.
I agree with the idea that the right time to decide how to handle AI is before it becomes extremely powerful. But he uses so much hyperbole, and what seems to me to be intentionally inflammatory language for purposes of creating fear.
It makes we want to take the cynical perspective that he's trying to cash in on fear to create a (valuable) following.
Re: Only You Can Stop an AI Apocalypse
#34I realize this may be satire but it is deeply relatable:
> AI will make critical decisions that we cannot understand.
Am I the only one that experiences this?
For example, you know the way, the road signs are telling you that you are on the right route, but your map application tells you to go another way, and you do it because you assume it is smarter or has more information than you.
You are analyzing a chess position and the engine suggests a move that you can't understand. The engine is far better than any human player ever, so you go with the engine's top choice the next time you reach that position.
You are betting on college basketball and everything you know about the matchup suggests team A should cover the spread against team B, but your model favors team B. Your model has performed much better than the human oddsmakers, so you ignore your intuition and bet on team B.
Re: Only You Can Stop an AI Apocalypse
#35Today I got a peek into the future. I asked ChatGPT to be the brain for a simple thermostat. I typed the following: "Imagine you are a thermostat for a house HVAC system. The commands you can use to control the HVAC are "HEAT ON", "HEAT OFF", "AC ON", and "AC OFF". I will tell you the temperature and my preferred temperature and you will command the HVAC system. Ready?" ChatGPT happily obliged. I used plain text to t…
Why is everyone bringing examples that are totally explained by how the tokenizer works. There is no intelligence it's Markov chains on crack.
Re: Only You Can Stop an AI Apocalypse
#36"Most AI researchers think good outcomes are more likely. This seems just blind faith, though" Or, you know, they might actually know something about the subject? On a similar note, most physicists don't think particle colliders will cause mini black holes which will swallow the Earth, but the fear gets repeated by non-physicists.
You are fallaciously appealing to (assumed) authority and missing entirely the fact that the author is a PhD in AI. This critically underscores why blind faith in authority is bad, and why people should do some basic research such as reading the article. By reading the article you would have understood the arguments the author is making on their merits, rather than on blind faith in imagined authority figures you are…
There are also physicist that believed cern will open a portal to hell... or just listen to the congress hearings about the Superconducting Super Collier in Waxahachie.
Re: Only You Can Stop an AI Apocalypse
#37Was there any reasonable way to prevent e.g. the current revolt against generative art software without first releasing the software so people could realize how powerful the tech was? Is there ever going to be a point where the tech we have is declared "good enough" so future conflicts like these are avoided?
There are good arguments that generative tech is setting everyone's expectations so companies will be forced to adopt AI (with all the implications attached) to keep up with the competition or go under, that it's dehumanizing creation through a vague equivalence of "intelligence" to "writing some words in a prompt", that it has the potential to devalue art and make the masses believe human artists are just really sophisticated prompt-replicators through buzzwords and misleading branding. And it's almost impossible to change people's minds once they're set in a position.
At the same time, the work on diffusion networks is rapidly accelerating, with novel techniques released seemingly every week, and researchers seem content with writing a couple of ethics paragraphs to go along with their full replicable source code release. It seems that imagination was not elaborate enough to foresee the effects their release would have on public discourse. Even with that discourse heating up by the day, the research continues.
It's never worked to just tell random people not to torrent things, even if it's illegal. They'll just do it anyway. Likewise, it's futile to tell people not to use software for bad, no matter how many legal clauses are attached. If the technology is available, it will be done at some point.
I'm convinced the only way this will be abated is if AI research were treated like gain-of-function research where the threat is not to human life but to individual, essential human qualities, like the way art is treated. That would mean any future technology would not exist, so it could not be exploited.
People in universities are inventing new software that nullifies the effects of the diffusion model training software because lots of people don't want the diffusion software to affect their lives.
I'm depressed about what this implies for the coming future of humanity. Is the train of endless technological progress just a given that must be followed through, consequences be damned? Is anyone in this sphere interested in stopping themselves and making do with what tech is already known? (Knowing that at least one person out of the billions on Earth is bound to say "no", rendering the exercise pointless.)
Re: Only You Can Stop an AI Apocalypse
#38Interesting but there are some red flags coming up here, like:
> This seems just blind faith, though.
This is logic? Eh, better to show your work. Otherwise this kind of wording seems like a hand wave.
Why does it seem blind faith--by what logic is it blind, and by what logic is it faith?
The blind-faithers in question are people you said were researchers after all, are probably OK being held to account.
Let's hear their response to being told they are relying on blind faith, as compared to let's say lots of experience in the topic and basic knowledge of probability. But that's just one example.
> A majority surveyed also acknowledge that utter catastrophe is quite possible.
Uh, the "quite" there is leading to a bit of a weasel-word effect. Where was "quite possible" derived from, vs. just "possible"?
There's a cumulative effect of being rushed to judgment here, so I hope there aren't a lot of important little details being rushed over too. The emotional floodgates can be hard to keep regulated in such a situation.
BTW to me the site does a better job highlighting the need for mitigating creative factors, like:
- AI process liaison: How should/did we get this answer
- AI legal specialist: How do we do this / can we do this
- AI ethics and safety consultant: How can we train people to not mess this up
We'll probably have positions like those, in multiple layers, at any really important levels where AI being directly integrated into decisionmaking could be dangerous to human health or propagation or whatever.
Add high levels of education, licensing, and certification as needed.
Otherwise it's a lot like complaining about anything else that's broadly capable, but doesn't have proper safeguards developed yet.
There's a comment here about the site being satire, but really--if so, it's extremely cryptic satire, to the degree that it's more like the kind of thing you'd write if you wanted to be able to claim 'satire' OR 'not satire' later on, as events develop...I really doubt it though.
Re: Only You Can Stop an AI Apocalypse
#39I realize this may be satire but it is deeply relatable: > AI will make critical decisions that we cannot understand. Am I the only one that experiences this? For example, you know the way, the road signs are telling you that you are on the right route, but your map application tells you to go another way, and you do it because you assume it is smarter or has more information than you. You are analyzing a chess posit…
Personally, I just replace "AI" in all such fearmongering statements with "man".
For example: "Man will make critical decisions that we cannot understand."
Re: Only You Can Stop an AI Apocalypse
#40Today I got a peek into the future. I asked ChatGPT to be the brain for a simple thermostat. I typed the following: "Imagine you are a thermostat for a house HVAC system. The commands you can use to control the HVAC are "HEAT ON", "HEAT OFF", "AC ON", and "AC OFF". I will tell you the temperature and my preferred temperature and you will command the HVAC system. Ready?" ChatGPT happily obliged. I used plain text to t…
Why is everyone bringing examples that are totally explained by how the tokenizer works. There is no intelligence it's Markov chains on crack.
Intelligence isn't really the interesting part. People are excited because we now have tools that interact well with ordinary human expression.