Search Ideas

Re consciously deciding to do something: once you’ve automatized some behavior, it’s hard to undo it just by virtue of being automatized, not necessarily because of entrenchment.

The trouble with ‘consciously deciding’ to do something in any case is that the conscious parts of your mind may be on board but other parts may not. But that discrepancy itself need not be entrenched.

Re consciously deciding to do something: once you’ve automatized some behavior, it’s hard to undo it just by virtue of being automatized, not necessarily because of entrenchment.

If you feel bad when you force yourself to stop doing something, you might feel bad because of the force, not because of the habit. My guess is they’re thinking more in terms of static memes.

The part about entrenched habits gets pretty close, though it doesn’t say much about the nature of the entrenchment or how to solve it.

I just noticed that the old TCS glossary has an entry on entrenchment and entrenched habits:

Entrenched ideas are ideas that you are unable to abandon even when they fail to survive rational criticism in your mind.

An entrenched habit is something that you can't stop doing even if you consciously decide to, or which makes you feel bad when you consciously force yourself to stop doing it.

I’ve looked at the glossary many times over the years, so maybe the seeds of my ideas about addiction came from it.

I just noticed that the old TCS glossary has an entry on entrenchment and entrenched habits:

Entrenched ideas are ideas that you are unable to abandon even when they fail to survive rational criticism in your mind.

An entrenched habit is something that you can't stop doing even if you consciously decide to, or which makes you feel bad when you consciously force yourself to stop doing it.

This is pretty cool! I think the part about entrenched habits gets pretty close, though it doesn’t say much about the nature of the entrenchment or how to solve it. Also, if you feel bad when you force yourself to stop doing something, you might feel bad because of the force, not because of the habit. My guess is they’re thinking more in terms of static memes.

Re consciously deciding to do something: once you’ve automatized some behavior, it’s hard to undo it just by virtue of being automatized, not necessarily because of entrenchment. The trouble with ‘consciously deciding’ to do something in any case is that the conscious parts of your mind may be on board but other parts may not. But that discrepancy itself need not be entrenched.

All that said, I’ve looked at the glossary many times over the years, so it’s definitely possible the seeds of my ideas about addiction came from it.

By the latter standard, neither nature nor random number generators are people, which is sensible; nor can nature create any given possible knowledge tractably -- this is true because the fact that all possible knowledge exists is only by way of the multiverse, which is a process that cannot be simulated in its entirety, even by a quantum computer, never mind tractability.

By this standard, a random number generator has universal creativity as well, and is therefore a person. So there must be a standard for personhood other than: able to generate any possible explanation. Such as: can do that tractably.

By this standard, a random number generator has universal creativity as well, and is therefore a person. So there must be a standard for personhood other than: able to generate any possible explanation. Such as: can do that tractably.

By this standard, a random number generator has universal creativity as well, and is therefore a person. So there must be standard for personhood other than: able to generate any possible explanation. Such as: can do that tractably.

Nature does have universal creativity; it can generate any possible knowledge. And all possible knowledge exists somewhere in reality.

But nature created genetic knowledge from nothing. So this is an example of something which does not have universal creativity which created knowledge ex nihilo.

This also admits of the distinction between AI and AGI (and "universal creativity") as being whether the system is capable of creating knowledge ex nihilo, as argued by Deutsch. Only universal creativity could create knowledge from nothing. Bounded creativity must start with something.

I keep returning to the notion of the space or domain in which simulated evolution so far operates in. It seems like we can say that current sim'd evolution can discover new knowledge via conjecture and criticism, but it is always bound by a domain predefined by fitness functions, automatic evaluators and so on, even if that domain itself contains many subdomains.

Then we can say that in nature, and in the minds of people, there is no externally defined space in which exploration is happening; the space is also evolving, also subject to criticism. I suspect this is part of how open-endedness comes about.

But the immediate question here was how to explain why AI is or is not "creative". Saying AIs are "narrowly creative" seems it could work, or saying they are creative within a fixed domain. The common intuition I think is that current AIs are "truly" creative, and I would say this is because the predefined domain (of LLMs, for instance) is gigantic, being sculpted by an internet-sized training corpus. But I suppose we should argue that "true creativity" means universal creativity.

I was curious if there are criticisms of the argument that current AI does legitimately create new knowledge.

This seems to me to be the same distinction that Deutsch and others have made between the genetic evolution we can simulate through evolutionary algorithms and the kind we actually observe in nature. I think it would be helpful to investigate evolutionary algorithms a bit further if you want to develop a clear distinction. This is how I describe it in my book:

There are several mechanisms that genes use to create variants, including sex, mutation, gene flow, and genetic drift, all of which appear to introduce change randomly. But we now know it cannot be entirely random. Something more is shaping what gets trialed, because when we model and simulate evolution using random changes, we never see the sort of novelties that arose in nature. We see optimization. We see exploitation. We see organisms become better at using resources they already use. But we never see a genuinely new use of a resource emerge. A fin may become better at swimming, but it does not become a limb. A metabolism may become more efficient, but it does not open up an entirely new biological pathway. And yet the natural world is full of exactly such extraordinary adaptations.

Move 37 was not explicitly present in the training data, nor designed by the programmers, and is extremely hard to vary (Deutsch's criterion for good explanations). Was the move present implicitly in the design of the system and/or the training data? Or inexplicitly? Does either of these mean the discovery of the move was non-creative?

Since evolution created genetic knowledge from nothing, it can be said to have the same "narrow creativity" as AI. The confusion over whether AI "is creative" can be resolved by saying that it is, but only narrowly (like evolution), and that the creativity defining people is universal, not limited to any domain. AI creates knowledge in domains it was designed for; AGI can create knowledge in all possible domains, each of which it designs itself.

AIs have created output that is not only novel, but seems to constitute new knowledge (resilient information), such as the famous Move 37 from AlphaGo. That is new knowledge because the move was not present in the training data explicitly, nor did the designers construct it.

Evidence that addiction and procrastination are related: https://www.quora.com/How-do-I-stop-procrastinating-and-addiction-to-the-internet
The person in that link asks, “How do I stop procrastinating and addiction to the internet?”

Of course it's true that a system confined to a given abstraction will only evolve theories of that scale. But a person can operate at all computable levels of abstraction. The growth of knowledge by people (e.g. Relativity) would only have happened if people can vary their abstractions arbitrarily, because Relativity solves no problems at any one given level of abstraction, but across many. Observer Theory might be right for certain systems, but is wrong for people.

Computational Universality only implies that all computable programs can be run by UCs. But what is relevant here is what programs can be reached by a given program -- synthesized by it. A UC with knowledge that only contains objectively whirlpool-scale conjectures (resulting from external stimulus or not) will not have niches relating to molecule-scale theories. Such theories solve no problems for it. So there will be no selection for those theories, so evolution will not develop them. Molecule-scale theories constitute intractable niches for the whirlpool system. They are still possible to run, if present, but that is not what's at issue. Observer Theory is correct if it is saying that the theories of reality developed by systems will depend on the abstraction level of their knowledge with respect to reality.

The aliasing that happens with the flipbook is a consequence of an imaging system. To suggest that theories/programs/explanations would be subject to aliasing in the same way suggests that they are derived from observation, which is Empiricism (false). They are created from mutation and criticism of existing knowledge, and this process can be performed by all universal computers. Any explanation/rendering/program runnable on one UC is runnable on all, so two observers can always converge to the same laws of physics.

Here’s a specific example of a working cure for addiction to sugar soda. Apparently, many people struggle with that.

Picture a man who likes the taste of soda but dislikes its fattening effects. (Sugar sodas are high in calories.) And let’s say the conflict between these two preferences is entrenched because his wife doesn’t want him to drink soda (for health reasons, say), and she chastises him, so he hides it from her and can’t talk about it openly, which makes error correction harder. In addition, he was made fun of as a kid for being overweight, so he feels awful whenever he thinks about dieting and can’t deal with the problem effectively. Such conditions are a breeding ground for entrenchment.

The solution in this case is to switch from regular sugar soda to the corresponding diet-soda equivalent. Like switching from Coke to Diet Coke or Coke Zero. It tastes virtually the same and has no calories. So now both parts of him get what they want. :)

This solution is especially interesting because it solves the addiction without having to make any major changes to behavior. A small, piecemeal change – switching from regular Coke to Diet Coke – removes the conflictedness and thus the addiction. More or less the same behavior can continue, showing that addiction is about conflicts between ideas, not about any specific or unilaterally unwanted behavior.

The key is to view even a small change to an idea as an entirely discrete, separate option. The small change can be enough to result in a common preference. The original preferences may get to live on the common preference as an approximation, as Popper might put it.

I'm realizing this is very related to Stephen Wolfram's "Observer Theory", which is interesting, but sounds worryingly relativist to me at times. Something like: Different observers will coarse-grain different laws of physics than the ones we have, for the same reason that the flipbook appears to have motion to us, but not to an observer viewing through a high-speed camera. Debating with LLMs about how that seems to violate computational universality has left me frustrated.

Better example maybe: A whirlpool in water only exists to an observer that can create whirlpools in its VR. If the observer only has molecule-scale abstraction, it cannot coarse-grain, so there are no whirlpools for it, or explanations in terms of them. (Such a system also cannot be a person, because a person can create all possible explanations).