Chapter 7 'Artificial Creativity' Part 1

[Gagesty]

"If you can't explain it simply, you don't understand it" - attributed to Albert Einstein. "What I cannot create, I don't understand" - attributed to Richard Feynman. "If you can't program it, you haven't understood it" - David Deutsch.

[Lance_Lough]

Great to see the next installment come up. I like to have the two DD books on my computer desk as I watch your presentations. My strategy is to read the chapter or section in question before your expansion—then go through it again after a few days of musing over it. I Always find something I'd missed or understood incompletely before. So, thanks again for your time and effort, Brett!

[majorhuman]

Hey Brett, love your content. Have you given some examples of what you’d need to see to consider or confirm an AI model is creating new knowledge or better explanations?

I’m looking for something testable ideally 😊

[jameskerr]

In the colour room thought experiment, doesn't the person who has "never seen blue before" have to know everything there is to know about the science of blue, in order to argue that subjective experiences are something over and above what can be described in purely physical terms?

But isn't Dennett's response to this that we don't really pay enough attention to thinking through the idea of "knowing everything there is to know about the science of blueness"?

After all, in Deutsch's terms, if a scientist (i.e., the person in the colour room) knew everything there was to know about blue, could they not already have built themselves a virtual reality colour-of-blue simulator and in that sense fully understand all possible experiences of seeing the colour blue? Would this not be identical to seeing blue or knowing exactly "what blue will look like"?

In which case, what is missing from the physcialist interpretation of experience? I think there is something missing, but why call it "qualia" and not simply "the mystery of my own existence as a conscious being"?

Related question: Does the idea that "subjectivity itself" (whatever that may be) has an objective reality even make sense?

I seem agree with everything that David D and yourself say (as far as I understand it) with the one exception of the qualia bit. And I really enjoy your excellent expositions of his magnificent ideas so thank you!

[neilhudson]

Thanks for your explanations of David's book! A couple of thoughts/observations: let us say that someone has programmed an AGI, such that the explanation for it being able to pass the traditional Turing test is that it is "the program wot dun it". Why though would such an AGI want to subject itself to the Turing test? Why should it consent to it-it knows it is intelligent after all! The answer would surely be that it wants to learn from the human "judge", things that it does not yet know, or that the human "judge" wants to learn things from it. Put differently, the traditional Turing test presupposes that the sole knowledgable authority is the human judge, as opposed to there being a dialogue between two non-authoritarian, fallible AGIs, one human, one not, but both consenting to the interaction.

[dennishackethal]

Excellent video again!

I don’t have the book in front of me right now, though of course I’ve read it (this particular chapter more than once). Does David suggest that consciousness is necessary for creativity? It’s certainly not **sufficient** as there are animals that are conscious but not creative, but I don’t understand yet why creativity should **require** consciousness.

Or maybe that’s not what he’s saying and you can simply disregard this comment. :)

[stegemme]

are you getting the date of publishing for this book Brett. There are a couple of references you've made where the inference is not correct. Seems like you're confusing TFOR for TBOI for example around 5:20

[jr]

👍👍
(the like function only allows for 1 thumbs up.. Here I supply 2 because merit)

[faroukbul]

Being lost in thought is such a mediocre experience unless you are creating or playing. And free will cannot logically exist, it doesn't mean we should accept reductionism. And you loose the sense of "I" every time you watch a movie, you just don't realize, please se Derek Parfit's Reasons and Persons part 3. He shows that the self it's an incoherent concept.

[Human_Evolution-]

What do you think about Sean Carroll? Have you read The Big Picture? He also has a podcast.

[MRLJ]

If possible, could you explain how the statements, 'If you can't/haven't programmed it, you haven't understood it.' and 'In practice the author’s explanation would
always be at some emergent, abstract level. But that would not prevent it from being a good explanation. It would not have to account for the specific computational steps that composed a joke, just as the theory of evolution does not have to account for why every specific mutation succeeded or failed in the history of a given adaptation. It would just explain how it could happen, and why we should expect it to happen, given how the program works.' are not in conflict.

It seems to me that if an explanation of AGI could account for how it could happen in principle, then programming it would only serve the purpose of creating that AGI. Don't get me wrong, the prospect of creating an AGI is extremely exciting, but the explanatory knowledge of how it works would be embodied in the program implicitly, and running it would be a critical test. It seems that programming it is not required for understanding AGI, unless David has the constructor theoretic notion in mind. :)