Love this! It’s an idea I’ve tried to share with others too: AI skeptics are needed for a realistic understanding of what AI can and can’t do. I’ve especially appreciated Chollet’s grounded AI optimism.
Thanks for your thoughtful piece and kind words about my writings. I guess it's inevitable in this polarized domain that I will be labeled an "AI skeptic", but I prefer (perhaps optimistically) to think of myself as an AI scientist, one who is trying to investigate---in an open-minded way---how these systems work and how valid are different claims about them. That being said, the *job* of a scientist is to be skeptical and to test (rather than accept) claims, especially one's own.
100% agreed. The work you're doing is invaluable at a time when many are concerned with other tasks. Thank you for that! I also agree that "skeptic" sounds weirdly dismissive. That's why I felt compelled to write this post. I actually doubted changing the word, but finally decided to leave it. Anyway, "AI scientist" it is!
The problem as I see it is today's internet is not meant for any serious questioning, accountability journalism or deeper reflection. Not only is this behavior typically not rewarded, it's actually punished because the status quo want to keep their Monopoly power.
If anything journalism continues to decline and along with it things like AI skepticism. While it may be fashionable to be contrarian now on networks like bluesky or substack, only a small fragment of Internet users care. If anything in the AI age of generative AI products, surveillance capitalism and behavior modification are stronger than ever.
Great piece Alberto! May I suggest another binary which captures something essential in current AI discourse?
1. AI Prophets (both hardcore skeptics and evangelical optimists - they predict what will happen based upon past experience)
2. AI Explainers (who attempt to explain “what it is” rather than predict “what it will become” as their method to move the space forward.)
All 3 of the great thinkers you mention would fit in the latter. (From the pieces I’ve read - you would too.)
Also, it’s striking how many similar assumptions, beliefs and thinking patterns those in the Prophecy camp have (on either side - optimist or skeptic).
I've always called myself a skeptical optimist. Rooted in realism, Stoicism, and in line with the Stockdale Paradox ("Each day is a challenge but the future is good.") and Amara's Law ("We overestimate technology in the short term and underestimate it in the long term.")
Therefore, I do not think there are two camps (optimists vs skeptics) with the latter split into two groups. No, there are *three* camps: the hypers (utopians, optimists), the doomers (dystopians, skeptics), and in the middle the realists (skeptical optimists).
Science is a tool, and honest debate can be swept aside if the tool master decides to cut funding or kneecap careers. Talking about AI skeptics is perhaps a bit misleading because the real issue is who is using the tools and why. I’m not skeptical of AI I’m skeptical of monopoly capitalism.
It is a nice article overall but it misses a crucial camp of skeptics that have argued against the very idea that something akin to human intelligence/mind can be replicated as digital computation. Namely people like Noam Chomsky, Roger Penrose, Joseph Weizenbaum just to name a few.
Yeah, but I don't think they count as AI skeptics. They're skeptical of something else, namely, the computational theory of the mind. (Still, the taxonomy I chose for this article is limited and surely non-exhaustive.)
Well it is hard to talk of meaningful AI without assuming some form computational theory of the mind and these authors also have explicitly argued against strong/human level AI.
Then I will just say that I can't possibly enumerate all the AI skeptics in the world haha I chose the three I consider most relevant today (Weizenbaum is dead and Chomsky/Penrose are 90+ year olds; let's give some space to the younger generations!!)
The age of arguments hardly matters on the contrary one should look whether they stood the test of time. But there is also a categorical difference between Mitchell, Chollet and Naranayan and the names mentioned namely that their critique is limited to the current state of AI and the hype surrounding it. For instance Chollet is more similar in spirit to Garry Marcus and others in that he thinks he has identified the magic missing piece for achieving truly intelligent AI - his challenge is called ARC-AGI for a reason. Whereas Chomsky, Penrose and Weizenbaum question whether trying to build intelligent computers that rival human mental faculties is a sensible endeavour/possible in principle in the first place.
I feel his blog overall is, practically, an essential counterweight to typical AI/LLM etc discourse...
However...I feel it's hard to actually determine if the AI 'cult' is literally anything other than just kids and/or conflicts of interest.
I.e., if the *primary* driver behind the 'cult' is actually just social media algorithms driving intentionally-hyperbole tweets from literal children and/or CEOs, is it *really* an actual cult? Or more of a pseudocult?
I dislike Ed Zitron, I think he's a good example of the other camp of skeptics (although I'd admit there are further subdivisions within that camp - he's not the worst).
Love this! It’s an idea I’ve tried to share with others too: AI skeptics are needed for a realistic understanding of what AI can and can’t do. I’ve especially appreciated Chollet’s grounded AI optimism.
Gotta love the skeptical optimists!!
Thanks for your thoughtful piece and kind words about my writings. I guess it's inevitable in this polarized domain that I will be labeled an "AI skeptic", but I prefer (perhaps optimistically) to think of myself as an AI scientist, one who is trying to investigate---in an open-minded way---how these systems work and how valid are different claims about them. That being said, the *job* of a scientist is to be skeptical and to test (rather than accept) claims, especially one's own.
100% agreed. The work you're doing is invaluable at a time when many are concerned with other tasks. Thank you for that! I also agree that "skeptic" sounds weirdly dismissive. That's why I felt compelled to write this post. I actually doubted changing the word, but finally decided to leave it. Anyway, "AI scientist" it is!
The problem as I see it is today's internet is not meant for any serious questioning, accountability journalism or deeper reflection. Not only is this behavior typically not rewarded, it's actually punished because the status quo want to keep their Monopoly power.
If anything journalism continues to decline and along with it things like AI skepticism. While it may be fashionable to be contrarian now on networks like bluesky or substack, only a small fragment of Internet users care. If anything in the AI age of generative AI products, surveillance capitalism and behavior modification are stronger than ever.
Great piece Alberto! May I suggest another binary which captures something essential in current AI discourse?
1. AI Prophets (both hardcore skeptics and evangelical optimists - they predict what will happen based upon past experience)
2. AI Explainers (who attempt to explain “what it is” rather than predict “what it will become” as their method to move the space forward.)
All 3 of the great thinkers you mention would fit in the latter. (From the pieces I’ve read - you would too.)
Also, it’s striking how many similar assumptions, beliefs and thinking patterns those in the Prophecy camp have (on either side - optimist or skeptic).
I've always called myself a skeptical optimist. Rooted in realism, Stoicism, and in line with the Stockdale Paradox ("Each day is a challenge but the future is good.") and Amara's Law ("We overestimate technology in the short term and underestimate it in the long term.")
Therefore, I do not think there are two camps (optimists vs skeptics) with the latter split into two groups. No, there are *three* camps: the hypers (utopians, optimists), the doomers (dystopians, skeptics), and in the middle the realists (skeptical optimists).
Science is a tool, and honest debate can be swept aside if the tool master decides to cut funding or kneecap careers. Talking about AI skeptics is perhaps a bit misleading because the real issue is who is using the tools and why. I’m not skeptical of AI I’m skeptical of monopoly capitalism.
That's a different topic, also worth exploring. Will you write about that?
I just wrote a paragraph about it. What more do you want from me?
It is a nice article overall but it misses a crucial camp of skeptics that have argued against the very idea that something akin to human intelligence/mind can be replicated as digital computation. Namely people like Noam Chomsky, Roger Penrose, Joseph Weizenbaum just to name a few.
Yeah, but I don't think they count as AI skeptics. They're skeptical of something else, namely, the computational theory of the mind. (Still, the taxonomy I chose for this article is limited and surely non-exhaustive.)
Well it is hard to talk of meaningful AI without assuming some form computational theory of the mind and these authors also have explicitly argued against strong/human level AI.
Then I will just say that I can't possibly enumerate all the AI skeptics in the world haha I chose the three I consider most relevant today (Weizenbaum is dead and Chomsky/Penrose are 90+ year olds; let's give some space to the younger generations!!)
The age of arguments hardly matters on the contrary one should look whether they stood the test of time. But there is also a categorical difference between Mitchell, Chollet and Naranayan and the names mentioned namely that their critique is limited to the current state of AI and the hype surrounding it. For instance Chollet is more similar in spirit to Garry Marcus and others in that he thinks he has identified the magic missing piece for achieving truly intelligent AI - his challenge is called ARC-AGI for a reason. Whereas Chomsky, Penrose and Weizenbaum question whether trying to build intelligent computers that rival human mental faculties is a sensible endeavour/possible in principle in the first place.
Would be very intrigued as to your thoughts on Ed Zitrons: https://www.wheresyoured.at/reality-check/
I feel his blog overall is, practically, an essential counterweight to typical AI/LLM etc discourse...
However...I feel it's hard to actually determine if the AI 'cult' is literally anything other than just kids and/or conflicts of interest.
I.e., if the *primary* driver behind the 'cult' is actually just social media algorithms driving intentionally-hyperbole tweets from literal children and/or CEOs, is it *really* an actual cult? Or more of a pseudocult?
I dislike Ed Zitron, I think he's a good example of the other camp of skeptics (although I'd admit there are further subdivisions within that camp - he's not the worst).
That’s interesting. Why specifically aren’t you a fan of his?
The main reason: he doesn't propose building an alternative. He just wants to see the current thing dedtroyed. The secondary reason: he's too caustic.