Will xAI overtake OpenAI as the leader in AI research by the end of 2023?
Basic
30
Ṁ3957
resolved Jan 9
Resolved
NO

Resolution will depend on benchmark performances and the creator's judgement, which will take into account motor skills, computational efficiency, and subjective general intelligence. Expert opinion may also be considered if the creator deems them to be reasonably unbiased.

Get
Ṁ1,000
and
S3.00
Sort by:
predicted YES

I created a new market on this question with a term extending to the end of 2024. https://manifold.markets/TruthGPT/will-xai-overtake-openai-as-the-lea-f35e2f0bf471

predicted NO

Why would anyone who could work for OAI / DeepMind / GoogleBrain / etc work for Elon after the public shitshow that Twitter has been?

@PartiallyTyped Twitter has been far from a shitshow, DAUs have been trending upwards, and Greg Yang is who I have faith in. Not Musk.

predicted NO

@JamieLegg Musk literally said they are cash flow negative, and 50% of advertiser revenue is gone. They still haven’t dealt with the consequences of GDPR violations and the other lawsuits.

predicted YES

@PartiallyTyped Revenue has begun being shared with creators on the platform as of 2 days ago which has seen sharp spikes in interest, combined with issues with Bluesky, Twitter will be fine. Either way, I trust the wider team at xAI and they have said they plan to ship products in weeks.

No shot, but worried about impartially based on comments so I will hold off on betting.

@SneakySly If xAI has no publicly available products or credible demonstrations of products, or if publicly available benchmark data does not definitively show that xAI is ahead, I will resolve NO.

@SneakySly Thank you for pointing this out; I sold my No shares after reading your comment and then reading others by the market creator. Probability of misresolution is too high.

If a third organization is the leader in AI research by the end of 2023 by your criterion, but xAI is ahead of OpenAI, how does this resolve?

@adele N/A.

Personally, I plan to vote "yes" once the market stabilizes; I'd never place my money (or mana) on anyone but Elon.

I'm worried you'll misresolve.

This is not happening. AI research isn't something where strong leadership providing a sense of direction is as useful as it is in 'building cars' or 'building rockets'. The latter two are primarily engineering challenges - very difficult ones, but ones following established principles and processes. In those cases, a single leader can make a significant difference by directing labor that would otherwise just be 'wasted' - say on the SLS or at Ford. When the car has a useless part and two teams each say the other team needs the part, Elon can just kill the part. Whereas in research, you don't know what to do, you don't know what you don't know, and you basically need a bunch of smart people to dance around in the darkness, try all sorts of things, look for insights. Compare, in general, the progress of 'science' to that of 'engineering', or the development of quantum mechanics to the Manhattan Project. Former - no single leader, many intelligent theorists doing work, many different experiments. Latter - massive engineering project, strong leadership. It's a difference of degree, not kind, leadership is still important for AI companies, but Elon's ability to 10x by setting an agenda is much lower. Also, Elon's not competing with lumbering established companies, but hot dynamic innovators.

Also, more importantly (but less interestingly), AI is just massively talent intensive, capital intensive, and just takes time. You need some of the smartest people in the world, a ton of capital (both of which Elon has) ... and time to train your XXXB parameter models (and not just once, but many times, you're doing research, you don't know what works!, (Elon doesn't have a time machine. Six months just isn't enough time to compete with established AI labs that have more talent than he does and are just older than X.

@jacksonpolack I think that Tesla's AI research will probably play a bigger role than you think. OpenAI's research has mostly--and completely recently--been focused on large language models trained on huge portions of the internet. Tesla is the undisputed leader in AI with the ability to behave in a 3D, physical environment.

This wouldn't mean much, except that vision is by far the highest-bandwidth channel of information. Even for humans. I strongly suspect that Tesla's prior knowledge (and GPU rigs), combined with the inherent advantages in building real-life, RL-based models, will dramatically tilt the race in xAI's favor.

Also, xAI's already announced a timeline for an update on their first product. I'm taking this as a sign that maybe some of the critical technical work has already been done, and the official launch of xAI was perhaps belated.

Last, and probably least, I just can't help but feel inclined to place my bets against the company that has dozens of employees working on vapid diversity, equity, and inclusion initiatives, and is well-known for creating language models that refuse to make predictions, hold opinions, decide anything, or say anything remotely politically incorrect.

If Elon Musk was capable of building a good AI research org then Tesla AI would be better.

© Manifold Markets, Inc.Terms + Mana-only TermsPrivacyRules