Well said throughout. I love that you are invoking Hegel here. We underrate how useful the big philosophical theories are to the current moment. Big transformational technologies (or social movements that believe the former, somewhat inconsequential to the point) deserve old, clear thinking.
Glad to hear about the initial ruling in anthropic's favor. What the judge said has been my perspective all along. It's no different than the way humans distill knowledge from books. As for possibly still being guilty of piracy even though they purchased the same books later, I think they're probably does need to be a sufficient enough penalty for that behavior to discourage it in the future even though they're active purchasing the books did bring them into compliance retroactively. The problem with there not being a sufficient penalty for the piracy would mean that companies would just go forward without even bothering to purchase the books. Obviously it doesn't look good for the companies to do this, but could their images be any worse at this point? The crux of the matter is what constitutes a sufficient penalty, especially when you're talking about a company with pockets as deep as Google. Of course, that's an ongoing issue in litigation, where the fines that pharmaceutical or chemical companies have to pay are large by normal human standards but a minute fraction of their profits.
As for the pollution of things like crochet designs, can you really blame that on AI? That's bad behavior of humans attempting to game the system in the way they've been doing since the early days of the Internet. I remember being hired by a company back in the early 2000s to write slop that contained certain keywords that would drive traffic from Google. Like repeatedly mentioning Britney Spears in a wide range of articles that had little or nothing to do with her (if anyone actually remembers who she is at this point). After writing one such article, I was disgusted with myself and never did it again. However, it was an earlier example of human greed making the Internet temporarily unusable. I think there's a fairly simple human solution to the problem Romero described. It's so obvious, I'm not even going to bother to elaborate.
I just don't understand why people devote so much energy to whining about the inevitable. There's literally nothing anybody can do to stop this juggernaut. Every ounce of their energy should be devoted to making the best of the situation. It's like Andrew Yang once said when asked about his best recommendation for climate change. He said move to higher ground. I commend the author for taking the high road. Although, I suppose what I am calling whining is the antithesis, and it was a necessary stage to reach the synthesis. I guess what bothers me is that it seems like we're well past peak antithesis and it's time for everyone to focus on synthesis.
I think you are arguing a philosophy of progress that's closer to Karl Popper than Hegel.
Here's my pantheon of the GOATs of progress, with a summary on what AI is good at:
Thales
Progress through reasoning from intuited first principles.
Plato
Progress through grasping the hidden, perfect mathematical order behind appearances.
Aristotle
Progress through systematic observation, classification, and reasoning about causes.
Knowledge needs framework and structure — and the question "what for?": his four causes, including the purpose toward which a thing is directed, gave inquiry its scaffolding and its sense of why things are as they are.
Archimedes
Progress through the sudden epiphany, disciplined into proof.
Copernicus
Progress through the paradigm shift: the productive reframe zig against the consensus frame zag.
Brahe
Progress through better instruments and the data they make possible.
Francis Bacon
Progress through patient, incremental, systematic method.
He reversed Aristotle's arrow — building upward from gathered particulars to general axioms rather than deducing downward from assumed premises — and recast knowledge as power, for "the relief of man's estate," while inventing method as a discipline to restrain the mind from fooling itself.
Kepler
Progress through the conceptual leap forced by stubborn data (Brahe's, BTW).
Kant
Progress through recognizing that the mind is not a passive mirror but an active shaper — it imposes the very structures, space, time, and causality, through which any experience becomes possible.
Hegel
Progress through the dialectic: contradiction resolved at a higher level.
Popper
Progress through conjecture and refutation, but NOT in Kantian necessity or Hegelian inevitability fashions.
Popper builds on Kant by keeping the active, framework-imposing mind — observation is always theory-laden — but converts Kant's necessary, certain a priori categories into fallible conjectures: namely, where an idea comes from is irrelevant, what counts is whether it makes risky predictions and survives the attempt to kill it. Thus, no Hegelian final truth is ever reached, only error endlessly eliminated.
Kuhn
Progress through paradigms punctuated by revolutions, not steady refutation.
He noted that Bacon mistook science's messy fact-gathering infancy for its whole method, missing that maturity means acquiring a paradigm that tells you which facts matter.
Deutsch
Progress through good explanations that cannot be tweaked without breaking.
The last three are really arguing over the whole list:
- Popper flattens everyone before him: it doesn't matter how an idea arrives, only whether it survives refutation.
- Kuhn pushes back — scientists mostly don't refute, they puzzle-solve largely in Baconian fashion inside a paradigm until anomalies force a revolution.
- Deutsch resolves Popper and Kuhn: the real engine is explanations that are hard to vary, which makes progress not just provisional but genuine, objective, and unbounded: "Problems are soluble," i.e., they're subject to being dissolved or loosened, not solved in a final epistemic sense.
My takes:
1) In Popperian fashion, AI doesn't care about paradigms;
2) In Deutschian fashion, AI makes a wider range of problems soluble;
3) In Keplerian fashion, AI can detect patterns within vast reams of data, as with the recent solutions of Erdős Problems...;
4) ...but in Kuhnian fashion, AI still needs a "Terence Tao" to verify that AI's reframes of the paradigm(s) in play are correct;
5) In Baconian fashion, humans can help AI make these type of progress by incrementally improving harnesses that constrain AI from fooling itself
So anyone working in AI or investing in AI can likewise use or bet on these 5 broad categories of progress.
Well said throughout. I love that you are invoking Hegel here. We underrate how useful the big philosophical theories are to the current moment. Big transformational technologies (or social movements that believe the former, somewhat inconsequential to the point) deserve old, clear thinking.
Yep. All trends are ultimately human trends
Glad to hear about the initial ruling in anthropic's favor. What the judge said has been my perspective all along. It's no different than the way humans distill knowledge from books. As for possibly still being guilty of piracy even though they purchased the same books later, I think they're probably does need to be a sufficient enough penalty for that behavior to discourage it in the future even though they're active purchasing the books did bring them into compliance retroactively. The problem with there not being a sufficient penalty for the piracy would mean that companies would just go forward without even bothering to purchase the books. Obviously it doesn't look good for the companies to do this, but could their images be any worse at this point? The crux of the matter is what constitutes a sufficient penalty, especially when you're talking about a company with pockets as deep as Google. Of course, that's an ongoing issue in litigation, where the fines that pharmaceutical or chemical companies have to pay are large by normal human standards but a minute fraction of their profits.
As for the pollution of things like crochet designs, can you really blame that on AI? That's bad behavior of humans attempting to game the system in the way they've been doing since the early days of the Internet. I remember being hired by a company back in the early 2000s to write slop that contained certain keywords that would drive traffic from Google. Like repeatedly mentioning Britney Spears in a wide range of articles that had little or nothing to do with her (if anyone actually remembers who she is at this point). After writing one such article, I was disgusted with myself and never did it again. However, it was an earlier example of human greed making the Internet temporarily unusable. I think there's a fairly simple human solution to the problem Romero described. It's so obvious, I'm not even going to bother to elaborate.
I just don't understand why people devote so much energy to whining about the inevitable. There's literally nothing anybody can do to stop this juggernaut. Every ounce of their energy should be devoted to making the best of the situation. It's like Andrew Yang once said when asked about his best recommendation for climate change. He said move to higher ground. I commend the author for taking the high road. Although, I suppose what I am calling whining is the antithesis, and it was a necessary stage to reach the synthesis. I guess what bothers me is that it seems like we're well past peak antithesis and it's time for everyone to focus on synthesis.
I think you are arguing a philosophy of progress that's closer to Karl Popper than Hegel.
Here's my pantheon of the GOATs of progress, with a summary on what AI is good at:
Thales
Progress through reasoning from intuited first principles.
Plato
Progress through grasping the hidden, perfect mathematical order behind appearances.
Aristotle
Progress through systematic observation, classification, and reasoning about causes.
Knowledge needs framework and structure — and the question "what for?": his four causes, including the purpose toward which a thing is directed, gave inquiry its scaffolding and its sense of why things are as they are.
Archimedes
Progress through the sudden epiphany, disciplined into proof.
Copernicus
Progress through the paradigm shift: the productive reframe zig against the consensus frame zag.
Brahe
Progress through better instruments and the data they make possible.
Francis Bacon
Progress through patient, incremental, systematic method.
He reversed Aristotle's arrow — building upward from gathered particulars to general axioms rather than deducing downward from assumed premises — and recast knowledge as power, for "the relief of man's estate," while inventing method as a discipline to restrain the mind from fooling itself.
Kepler
Progress through the conceptual leap forced by stubborn data (Brahe's, BTW).
Kant
Progress through recognizing that the mind is not a passive mirror but an active shaper — it imposes the very structures, space, time, and causality, through which any experience becomes possible.
Hegel
Progress through the dialectic: contradiction resolved at a higher level.
Popper
Progress through conjecture and refutation, but NOT in Kantian necessity or Hegelian inevitability fashions.
Popper builds on Kant by keeping the active, framework-imposing mind — observation is always theory-laden — but converts Kant's necessary, certain a priori categories into fallible conjectures: namely, where an idea comes from is irrelevant, what counts is whether it makes risky predictions and survives the attempt to kill it. Thus, no Hegelian final truth is ever reached, only error endlessly eliminated.
Kuhn
Progress through paradigms punctuated by revolutions, not steady refutation.
He noted that Bacon mistook science's messy fact-gathering infancy for its whole method, missing that maturity means acquiring a paradigm that tells you which facts matter.
Deutsch
Progress through good explanations that cannot be tweaked without breaking.
The last three are really arguing over the whole list:
- Popper flattens everyone before him: it doesn't matter how an idea arrives, only whether it survives refutation.
- Kuhn pushes back — scientists mostly don't refute, they puzzle-solve largely in Baconian fashion inside a paradigm until anomalies force a revolution.
- Deutsch resolves Popper and Kuhn: the real engine is explanations that are hard to vary, which makes progress not just provisional but genuine, objective, and unbounded: "Problems are soluble," i.e., they're subject to being dissolved or loosened, not solved in a final epistemic sense.
My takes:
1) In Popperian fashion, AI doesn't care about paradigms;
2) In Deutschian fashion, AI makes a wider range of problems soluble;
3) In Keplerian fashion, AI can detect patterns within vast reams of data, as with the recent solutions of Erdős Problems...;
4) ...but in Kuhnian fashion, AI still needs a "Terence Tao" to verify that AI's reframes of the paradigm(s) in play are correct;
5) In Baconian fashion, humans can help AI make these type of progress by incrementally improving harnesses that constrain AI from fooling itself
So anyone working in AI or investing in AI can likewise use or bet on these 5 broad categories of progress.
Wow. You lost me at hello!