This market is intended to use the same resolution criteria as https://manifold.markets/dreev/will-ai-have-a-sudden-trillion-doll, but for 2025 instead of 2023. For the sake of consistency with that market, I'll ask @dreev for a judgement if the resolution seems non-obvious to me. For convenience, what follows is the description of the linked market:
I'm picking "trillion+ dollar impact" as a proxy for "obviously life-changing for normal people". It need not count as human-level AI aka artificial general intelligence (AGI). Examples of things that would surely count:
Virtual assistants that are better than well-paid humans
Superhuman art, i.e., people tend to prefer to read / view / listen to AI-generated art
AI generating wholly new science/tech
A technological singularity, obviously
Level 5 self-driving cars or level 4 available mostly everywhere
Nov 29, 5:10pm: Will AI have a sudden trillion+ dollar impact by the end of 2025? → Will AI have a trillion+ dollar impact by the end of 2025?
1,000People are also trading
My tentative opinion on this is that this didn't happen by the deadline. If you sum up the valuations of relevant companies, they did make it past a trillion dollars by the end of 2025 but of course that's the market's expectation of the net present value of their future profits. The original market was clear that we're measuring this by current impact on normal humans. If we go through the list of examples of things that would "surely count":
Superhuman virtual assistants -- NO, this is waiting for more capable agents (aka computer use) at least
Superhuman art -- This was about what people prefer to watch / look at / read and I know that in blind tests AI does well, but, I don't know. People aren't paying to see AI-generated movies or read AI-generated books. Definitely some ambiguity on this one. It doesn't feel like a trillion-dollar impact in itself at least.
AI-generated science/tech -- The "wholly new" part of this inclines me towards NO.
Level 4 self-driving cars ubiquitous -- NO, still just a handful of cities.
Level 5 self-driving cars -- Full NO.
Overall, for "obviously life-changing for normal people" I would say no, at least not obviously, not quite yet.
Superhuman art; people tend to prefer to read / view / listen to AI-generated art
For comparison, chatgpt says the following industries are:
Gaming: 500 billion
Porn (what it called "sextech"): 40 billion, but hard to estimate, it said
Movies: 300 billion
Sports: 500 billion (surprises me it's more than movies)
Drinks: 3.5 trillion
So one way we can think of this prediction as being something like "as economically important as sports + gaming by end of 2025". This makes me update away from my YES buy. I'd buy YES for 100-500 billion by end of 2025.
@EliezerYudkowsky I think that phrasing was taken from my version of this market and the word "sudden" isn't adding anything. I guess it seemed more natural in my version since it's specifically about the impact happening by the following calendar year so no such thing as happening gradually in that timescale.
We can presumably say the same for this market. I agree it's confusing.
(In case we fix the titles and this dialog makes no sense, the original was "Will AI have a sudden trillion+ dollar impact by the end of 2025?")
@dreev is correct. I will fix the title this afternoon unless someone gives a reason not to by then.
@dreev I believe the >$1t rise in Nvidia's valuatiom meets those criteria as it is a single company and clearly attributable to AI. Do you have thoughts?
@RobertCousineau The growth in Nvidia’s stock isn’t actual impact, it’s a prediction of future impact. If you look at the resolution criteria, the key seems to be having a meaningful impact on the average person’s life, which AI definitely hasn’t met yet.
@Gabrielle Sounds exactly right. I'd love to get this nailed down tighter, but that's the spirit of the question.