Hacker News Reader

Comment by s1k3s | Hacker News Reader

The demos are great. Could someone explain what’s in it for Meta open sourcing all these models?

A competitive opensource project basically destroys the pricing power of all closed-source alternatives.

If you're a company and wanted to integrate an LLM into your product if the choice is between several equally good models, but one is free and open-source which would you pick?

Aside from keeping competition at bay, this move also gives Meta leverage because ecosystems are now being built around their projects. If these models see wide-scale adoption they could later launch AudioCraft+ as a licensed version with some extra features for example.

Alternatively, they might offer support or hosting for their open source projects.

Right now though I think the primary benefit of these open sourced models is to attract talent. If Meta is seen as one of the leaders in AI then researchers will want to work for them simply for the prestige.

Arguably one of the reasons Meta has been behind so many awesome projects like PyTorch and React over the last decade was because they were seen as the cool place for recently graduated, but talented software engineers to work in ~2010.

They want to comoditize the offerings of OpenAI, Google, MS, and Apple. Also, they gain mindshare and good will after years of bad publicity. Some back contributions might help them improve the models for free.

If they just keep their models, people won't be interested and will build over ChatGPT or Bard.

Not a fan of Meta, but haven't they generally been pretty forward with open sourcing their tech?

Commoditize Your Complement?

https://gwern.net/complement

What is it a complement to though?

Meta has several of the biggest UGC platforms, and in this case the complement is content itself. Reels with autogenerated (and royalty free) background music is the obvious example but I'm sure there are more. Maybe creative for ads as well?

To Metaverse access. Filling the metaverse with engaging interactive 3D content is an insane job with 2020 technology. It requires a huge amount of a range of skilled labor to create 3D models, soundtracks, NPC dialog, visuals et.c. to make a compelling experience. In 2030 that may have been reduced to that everyone with creativity and Internet access can do it. Sure, most of it will be silly things - but so is social media today, does not make it any less of a commercial success. And there will be be millions of semi-pro creators to create the things with higher production value, like with videography today.

Content is a complement to social media.

In short term it's social media, because people will share whatever they generate on social media. But I don't think it's a very strong incentive to invest in AI for Meta.

If anything it makes them appear to be one of the best places to work at to do research. Could be them playing the long game.

The demos are, unsurprisingly, soulless muzak. This contributes nothing to our culture.

Was asking myself the same earlier, I'm sure it is largely to do with publicity and the fact that selling these services is not their core business. At the very least releasing this stuff probably won't damage their core business but will take the sheen off of some other big names.

I wondered though, generative AI is hurling us into a world where we'll need more mechanisms to sort real from fake, provenance will play a large part, and meta's platforms could be part of the answer. i.e. content linked to actual verifiable people.

Somewhat relevant, Yann LeCun insisted the research should be open sourced. At least in an academic sense.

He touches on it briefly in this podcast episode: https://www.therobotbrains.ai/who-is-yann-lecun

They will own most popular open models so they can dictate the direction in open source AI.

They haven't opened sourced much. Open models/closed weights restrictive non-commercial license is something I guess.

They are trying to kill the market before they get left out.

The same move that Microsoft did back in 90's to kill Netscape. Make your product the one available to masses, next generation of users will be using your product.

I was just thinking how Google made Android free to check Microsoft. This is Meta checking Google.

Checking OpenAI. Google is still playing checkers.

I just can't get how bad Google is doing. They have a ton of top researchers, papers, money, just no good LLMs. It's like OpenAI was first to the punch, and everyone else just saw $$$. Meta was smart to go down this open source road, as the masses will start training their llamas one way or another. Personally I believe the "intelligence" aspect will asymptote, so even having exclusive access to a "super AI" (i.e. hypothetical 1T parameter model like a GPT5) won't be that much of a step behind the lesser AIs, and as soon as you grant access to the masses they will start to use some transfer learning to make their "lesser" models better. AI applications though still need a lot of work. The models are smart or general purpose enough to be useful to the average person out of the box.

The problem also is that Google is making lot of grandiose announcements about tools and models that nobody can see nor use. This is a serious credibility problem in the long-term.

You can use some of them. They have an “AI powered” search (as if their previous search isn’t considered AI anymore). It’s an experiment you can turn on. For programming questions it’s not terrible.

That said, there are a ton of “look at this cool thing out research team did” and then you never hear about them again things from Google. They even built a music generator that was closed to the public until recently.

https://blog.google/technology/ai/musiclm-google-ai-test-kit...

The fact you believe, rightly or wrongly, that meta is ahead of google on ai explains why meta would open source this. It’s a good reputation to maintain.

Checking Google or OpenAI (or both?)

theoretically what's in it for them is that people will build content faster and with less barriers for eventual consumption on their platforms

If people love hanging out with chatgpt or bard, they won't be wasting their precious little eyeballs on FB/Insta

Reply on news.ycombinator.com