TheaterFire

Meta chief AI scientist Yann LeCun plans to exit to launch startup, FT reports

Posted by brown2green@reddit | LocalLLaMA | View on Reddit | 43 comments

Reply to Post

43 Comments

CYTR_@reddit

He has officially announced his departure. I can't help but feel a touch of chauvinism when I read this. With a bit of luck, the French government will invest in an attempt to bring one of its national talents back home. I think that beyond the financial means (a guy like LeCun would surely need 1 billion/y just for operations), It could very well continue its international partnerships and recruit locally/import engineers and PhDs. I think there's no shortage of people who could work with him. But between the lack of infrastructure (there is nothing in the EU that is comparable to the computing power of META) and a completely paralyzed public authority... 🥸
View on Reddit #71790517

uhuge@reddit

I hope it's something ambitious and transparent he's onto!
View on Reddit #71483910

FullOf_Bad_Ideas@reddit

I think it's a good move. I think he can create many good, useful architectures and models with only a tiny bit of CAPEX spend of Meta. His name is enough to attract investment that will give him a reasonable research team and access to compute. Some people laid off from FAIR team recently may also be able to find work there.
View on Reddit #70997497

Ankmeister@reddit

He was completely against LLMs. His posts are all over linkedin. And I tend to want to agree with the founder of neural networks used by all of these LLMs. He also believed we are still a breakthrough or two away from anything that resembles actual AGI. I think he will build a solid team then get acquired by OpenAI for billions assuming they have any of their negative billions left :p
View on Reddit #71080101

Formal_Drop526@reddit

>I think he will build a solid team then get acquired by OpenAI for billions assuming they have any of their negative billions left :p If the team share's Yann's philosophy, they would never join OpenAI.
View on Reddit #71187743

FullOf_Bad_Ideas@reddit

I think he was against LLMs being a path to superintelligence. And I think it's hard to not nod a little when you read his specific claims. So, it could be seen as being against LLMs, but it was a suggestion to not throw everything at LLMs to the point of suffocating everything else, just because LLMs are easy to scale. >I think he will build a solid team then get acquired by OpenAI for billions assuming they have any of their negative billions left :p I hope he won't get acquired, we need more independent teams.
View on Reddit #71081555

JadeSerpant@reddit

Losing Yann LeCun while paying a 24 year old $250M. Zuck 200 IQ move fast break things strategy. Let's see how it pans out.
View on Reddit #71055090

brown2green@reddit (OP)

I guess Meta is not going to open source anything useful any time soon. Yann LeCun has often stated that Meta's committment to open research was what made him accept to work for the company in the first place.
View on Reddit #70993544

YouDontSeemRight@reddit

META open sources a lot of useful AI. They just released a bunch of universal translation models and DinoV3 is in a class of its own in understanding images. It allowed Hunyuan to create Hunyuan 3D 2.1 for example.
View on Reddit #71000146

buppermint@reddit

All those things were released by FAIR (Lecun's org) that has been getting de-prioritized and de-resourced over the last year. The new AI org getting all the money (Meta Superintelligence Labs) is anti-open research so I expect you will see way less open source soon.
View on Reddit #71011898

matthewjc@reddit

Do you work at meta?
View on Reddit #71019456

pm_me_github_repos@reddit

Likely not. MSL includes FAIR. Yann reports to Alex Wang
View on Reddit #71027333

One-Employment3759@reddit

Ah yeah, having to report to Alex Wang would do it.
View on Reddit #71053917

Budget-Juggernaut-68@reddit

And they just open sourced a new ASR model that supports a lot of languages.
View on Reddit #71033519

YouDontSeemRight@reddit

Have you looked into it?
View on Reddit #71043793

Budget-Juggernaut-68@reddit

Just briefly. Have not tested it. It does claim to support a lot of languages and dialect. The paper is mostly "My model is better than the other models when compared to others". Will need some testing to see how well it does. Interesting paper though
View on Reddit #71048070

YouDontSeemRight@reddit

Yeah, really curious now. One day lol.
View on Reddit #71048498

Material_Policy6327@reddit

AI is way overloaded. It’s models
View on Reddit #71023642

brown2green@reddit (OP)

What I mean is that going forward Meta [their recently-founded Superintelligence Labs] will probably avoid publishing _commercially useful_ research (that could potentially give advantages to their competitors) or powerful generative models with direct downstream applications in competition with their services. DINOv3 (which was made by FAIR researchers) is a vision embedding model that needs applications built around it. It _can_ be useful, but it's not immediately useful.
View on Reddit #71014769

Corporate_Drone31@reddit

Let's realistically admit that when most people in this community mean by "useful AI", is "large language models". That's not to say I disagree - I'm quite happy they released that ASR model recently. It will probably replace Whisper soon enough.
View on Reddit #71006815

mtmttuan@reddit

AI is more than LLM.
View on Reddit #71004834

-p-e-w-@reddit

LLMs are 99% of AI though. Investors aren’t planning to pour half a trillion into AI in the next 12 months because of image classification models that can tell a dog from a cat.
View on Reddit #71011116

fallingdowndizzyvr@reddit

> LLMs are 99% of AI though. No it's not. You are forgetting about image/video/audio gen.
View on Reddit #71012382

-p-e-w-@reddit

Compare the valuation of LLM companies to that of image generation companies. Not even in the same league.
View on Reddit #71032361

fallingdowndizzyvr@reddit

And what does that have to do with how many people use them? But if you must. Alibaba 396B Tencent 749B OpenAI 8-500B depending on who you believe.
View on Reddit #71038937

FullOf_Bad_Ideas@reddit

>LLMs are 99% of AI though. I don't think so. Maybe in terms of compute spend (training only) they're about 50%. Video models are probably 30%, image models 10%. And the rest get the remaining 10% of compute. It looks completely different in terms of researcher time allocation than in terms of compute though. BADAS is a recently presented model based on V-JEPA2 architecture that allows for collision detection. I think it's a super cool model and it has quick path to revenue generation, much quicker than LLMs. https://x.com/getnexar/status/1980252154419179870?s=20 This model will be saving real lives very soon.
View on Reddit #71012391

Down_The_Rabbithole@reddit

Very ignorant thing to claim
View on Reddit #71012096

QuackerEnte@reddit

Which could mean: they aren't committed to open research anymore, so he literally has no reason to stay with Meta.. I hope I'm wrong here and just extrapolating from unrelated data.
View on Reddit #71010182

s101c@reddit

Can anyone explain what is Meta doing right now? Their AI chatbot in Whatsapp is based on Llama 4. With their megabudget, where are there any results post-spring 2025? Any?
View on Reddit #71005819

Clueless_Nooblet@reddit

Wishing him all the best. He's a bright guy, his talent was wasted at Meta.
View on Reddit #70995768

05032-MendicantBias@reddit

Zuck was giving him all the compute money could buy. It was a fair deal. Zuck seems to have realized one *need* artificial super intelligence to make the AI investment model work, and rather than realizing that the investment isn't going to pay anytime soon (possibly decades), he seems to have that as a goal. It was the same for the metaverse. It took like 40 billions to make a worse version of VR chat.
View on Reddit #71053032

autoencoder@reddit

Well, I think he got a fair bit of useful money from it, and he can pursue his dream more easily. I don't think it was wasted.
View on Reddit #70999392

05032-MendicantBias@reddit

Good move. Gotta tap into that venture capital money spigott while it's still gushing open.
View on Reddit #71052898

maggienoodle007@reddit

Heard the creator of PyTorch was also leaving company, interesting timing
View on Reddit #71047386

ab2377@reddit

since meta has zuck problem, this is great 👍
View on Reddit #71037411

AppearanceHeavy6724@reddit

Meta AI is fubar at this point.
View on Reddit #71003241

Question-31080@reddit

Hey Meta, GTFO.
View on Reddit #71016729

the__storm@reddit

Not surprised - from the outside it seems like Meta has dismantled FAIR and fired a big chunk of its former employees, and is basically restarting with a new organization less focused on research.
View on Reddit #71007787

a_beautiful_rhind@reddit

Couldn't take the wang.
View on Reddit #71006370

ASTRdeca@reddit

Interesting. I wonder what this means for the future of JEPA
View on Reddit #71001384

brown2green@reddit (OP)

Other groups are starting to work on JEPA-like systems already, so my guess is that he'll focus more on it, hopefully with proof of concepts that the general public will be more interested in. The [CALM paper from the other day](https://www.reddit.com/r/LocalLLaMA/comments/1opabzi/instead_of_predicting_one_token_at_a_time_calm/) obviously took ideas from it (latent space prediction, energy-based model) although it didn't cite LeCun or JEPA at all, as far as I could see.
View on Reddit #71002005

JawGBoi@reddit

Not surprised to be honest.
View on Reddit #71000770

john0201@reddit

He got his cash, now he can do whatever he wants. Not sure he ever really wanted to work there.
View on Reddit #70999356