Justin T. Sampson

Seasoned software engineer. Perennial grad student. Aspiring philosopher.

[Justin T. Sampson @sampsonetics]

This is me staking out a position in the human-vs.-A.I. consciousness debate. Much of this post was drafted a long time ago, so any similarity of topic to recent writings by others is a happy accident.

https://t.co/ocbjcyEYAB

[Johannes Kleiner @JohannesKleiner]

Our paper received its first criticism. 🎉 I'm super grateful to Justin Sampson @sampsonetics for engaging with our work so thoroughly!

Here is my reply, in case anyone is interested. The criticism is linked in the post.
👉 https://t.co/i82F4kVlne

[Justin T. Sampson @sampsonetics]

@JohannesKleiner I'll keep that in mind! We'll have to discuss the math some more if I'm going to formalize my response enough for publication. 🙂 I have another paper in progress but once that's closer to preprint stage this would be fun to work on. My Substack posts can be like pre-preprints!

[Justin T. Sampson @sampsonetics]

@JohannesKleiner Thanks for the diff! I just posted my response to your no-go paper on Substack. I hope my genuine appreciation for your work comes across despite the elements of disagreement. https://t.co/EPznTgR5py

[Justin T. Sampson @sampsonetics]

@JohannesKleiner Thanks for the diff! I just posted my response to your no-go paper on Substack. I hope my genuine appreciation for your work comes across despite the elements of disagreement. https://t.co/EPznTgR5py

[Justin T. Sampson @sampsonetics]

@JohannesKleiner @lollolsasa That is, those aspects of cognition that go beyond computation are not "uncomputable" per se, they are just contingent facts about our actual experiences, opinions, guesses, etc. We can't "know" any Gödel statements, but we can believe unprovable statements for pragmatic reasons.

[Justin T. Sampson @sampsonetics]

@JohannesKleiner @lollolsasa Hmm, you only mention Gödel briefly in your thesis, and you don't use incompleteness in your proofs, so I take this as a general question. I agree with your answer. The true-but-unprovable statements in Gödel's theorem aren't really useful facts that we care about knowing.

[Justin T. Sampson @sampsonetics]

@JohannesKleiner I've discussed that preprint with some of its authors in private but I don't fully understand the math yet so I will refrain from commenting on it further at the moment. 🙂

[Justin T. Sampson @sampsonetics]

@JohannesKleiner The IIT 4.0 macroing rules (from the preprint that I linked earlier in the thread) are formulated differently, so the background is not held "constant" per se, but driven by the same intuition about statistically isolating the causal influences of macro elements on one another.

[Justin T. Sampson @sampsonetics]

@JohannesKleiner Finally (for now), the IIT 3.0 rules, if you include black-boxing which came later, handled irreducibility of the macro units by partitioning the system at the micro level when calculating Phi. The IIT 4.0 rules focus instead on each unit alone rather than all the units together.

[Justin T. Sampson @sampsonetics]

@JohannesKleiner By the way, there's a preprint available for the IIT 4.0 macroing rules, which are only mentioned in passing by the main IIT 4.0 paper itself: https://t.co/ZXFD2FPN24

[David Chalmers @davidchalmers42]

@sampsonetics this is a superb MA thesis! i just skimmed most of it. it's just the sort of thing i was looking for, both explaining why certain official versions of IIT say von neumann simulations have low phi, and outlining an alternative version that doesn't have that consequence. your "snapshot IIT" is very much the sort of thing i had in mind and as you say it's very much in the spirit of causal structuralism. it's great to see all this worked out in detail.

[Justin T. Sampson @sampsonetics]

@JohannesKleiner @davidchalmers42 Thanks for the mention. :) My thesis reviews the argument about von Neumann memory units in IIT, and I conclude that IIT is not yet formalized enough to settle the question. I offer a new formalization that comes out in favor of consciousness in computers. https://t.co/ZamckaKk1N