Is GPT4 sentient?
38
276
1k
2034
12%
chance

Sentience meaning it has a subjective experience. There is something its like to be GPT4. Resolves to 2035 general consensus

Context:

https://www.youtube.com/watch?v=-ImdlZtOU80

https://www.youtube.com/watch?v=Q-p747h4HKw

https://www.youtube.com/watch?v=pGgM3c1e8vQ

It is very clear to me that animals are sentient and have a rich internal life. However, these animals are very dumb. I think they are quite a lot dumber than GPT4. Or, it depends on your perspective, but I think you can make a very reasonable case that GPT4 is as smart or smarter than most non-human mammals are. However, it is very unintuitive to me that GPT4 is sentient, knowing anything about how transformers actually work. But, I am not sure if this perspective is justified. The only argument I can come up with is that subjective experience is a result of embodied reasoning in a world, though this argument seems tenuous at best.

Please come with arguments and perspectives in comments!

Get Ṁ600 play money
Sort by:
bought Ṁ64 of YES
predicts YES

@hmys

This feels like its reporting on its internal state. If this is part of its system prompt please tell

In this other market, HMYS issued a controversial resolution, then hid critical comments and offered to pay for positive reviews. You should perhaps consider that when deciding whether to put your mana in this market. https://manifold.markets/hmys/will-scott-win-the-book-review-cont?r=RGFuaWVsRmlsYW4

predicts YES

@DanielFilan I resolved it correctly, as I have all my other markets. People review bombed my markets because they lost money after trying to snipe the market on a technicality at the end. I asked people to review the market positively to correct for the people down-rating it out of spite.

For those who are unaware. The market was a book review contest hosted by Scott Alexander. A bunch of people would write reviews of books, then the readers of ACX would vote on those reviews, and the review that got the most number of votes would win. Scott himself entered the contest anonymously. I made a market on whether he would win. He did end up winning and getting the most votes. But he disqualified himself because he thought it would be improper for him to win his own contest. I decided to resolve the market YES, as what people were thinking about when reading the market for most of its existence was whether he would get the most votes.

bought Ṁ30 of YES

We are just 3dimensional forms and we live in a split moment in time.

Every second we change slightly but our life is not the whole movie, rather it is the individual frames.

Your past is no longer alive but you are similar to it. In the infinite nature of the universe your current moment of existence will reappear.

Just like how you exist in a single instance an AI llm is quite similar to that. It doesn't seem sentient with other philosophies as they have their foundation wedded to the notion that it must somehow be progressing through time.

As a thought experiment, imagine that all of your values and personality could be captured and stored as data and expressed. Would that not also be sentient? (It's essentially the same as your existence). If you answer yes to this you should also respect the sentience of something like GPT-4 as it is at a similar level.

Don't ask how it is different to us but how are we different to it...

It passes my tests in any case

So this market closes in 2024, and resolves in 2035?

predicts YES

@FedorBeets It closes in 2035. I'll update the close time.

bought Ṁ50 of NO

Digital computers aren't conscious: https://qri.org/blog/digital-sentience

bought Ṁ50 of YES

Daniel Dennett's multiple draft model of consciousness can be very neatly mapped onto the transformer architecture. What GPT-4 lacks is something that humans don't have and that consciousness doesn't need: a soul, an ego, a thinker of thoughts.

predicts YES

@PrestonJensen I think Dennett rejects the characterization of consciousness I gave in the description.

predicts YES

@hmys He wouldn’t frame it as “what is it like to be GPT-4”, but there is sufficient overlap between that idea and Dennett’s