3 ms·

Meta is a great example of why you need a really good vision and execution. Imagine an alternative world where instead of acquiring Oculus and later pivoting t

Meta is a great example of why you need a really good vision and execution. Imagine an alternative world where instead of acquiring Oculus and later pivoting to the metaverse, Zuckerboard went all in on AI (LLM weren't a thing quite yet then), and instead rebranded the company Fabula (tale in latin), and developed AI technology to accelerate their original mission of connecting people. Though it’s true Facebook has an amazing AI team, it’s my understanding none of it has really been turned into product. In that regard they have been more like Google and less like OpenAI They likely would've created something similar to LLMs by themselves in short order, but in the worst case could have made their own LLM after the Google paper. You could imagine all sort of neat affordances given by an AI: - Friendship level suggestions on what to do - Meetup work, suggested on similar likes - Improved suggestions etc I bet that hypothetical pivot would've been better than the metaverse. At this point (2023) you might imagine things iterating to the point where users have some sort of assist that helps facilitate engagement with your facebook friends (with ads, of course, hah).

ripvanwinkle 4y ago
Part of the Meta's problem is this spiel about connecting people. Conversations internally are all about monetizing users and not losing any revenue due to new privacy regulations. Apart from the Metaverse which is also cloaked in "connecting people" there is no effort to take the technology and capabilities in the company and conquer new domains. Like Google getting into Office collaboration and Cloud computing etc. So that spiel about "connecting people" while being a bunch of whitewash is also restricting possibilities for the company
lostmsu 4y ago
> Zuckerboard went all in on AI Hindsight is 20/20
suzzer99 4y ago
The problem is I don't need any of this stuff and I know a ton of people who feel the same way. If FB's timeline had a chronological option like twitter, I would switch to it tomorrow, then mute people who bugged me. Every algorithm seems to end up optimizing for engagement, which basically means car wrecks, conspiracy theories, and other outrageous controversial stuff. Youtube will never stop recommending Jordan Peterson, no matter how many times you vote it down, because the algorithm has figured out that if it can get you into JP, it will open up a whole new world of engagement as you head further and further down the alt-right rabbit hole. So it's worth a periodic try at the hail mary, even for users who previously showed active aversion to JP.
pavlov 4y ago
Have you heard of LLaMa? Segment Anything? Those are just two major Meta AI projects released within the past two months. Generative AI is a major part of Zuck’s metaverse hopes and investment priorities.
- PLenz 4y ago
  To be fair, releasing the interesting parts of Llama was a decision made for instead of by them
- endisneigh 4y ago
  uh, what I'm saying is that all of that could've begun in 2014, not just now. clearly they're playing catch-up. Before OpenAI what language models did they release, exactly? how is segment anything playing into their metaverse strategy exactly? it's very clear that they're relasing this stuff to make it known that they're not going to fall behind to Microsoft (OpenAI) and Google. it clearly was not part of their original strategy around the metaverse, which is why it's all coming out now.
  - ianbutler 4y ago
    https://ai.facebook.com/blog/roberta-an-optimized-method-for-pretraining-self-supervised-nlp-systems/ https://ai.facebook.com/blog/roberta-an-optimized-method-for... As just one example, which was (is?) one of the most used flavors of transformers from the BERT era.
    - endisneigh 4y ago
      I'm not talking about papers, I'm talking about a product offering, ala ChatGPT. Even Segment Anything isn't a real product. It's a glorified tech demo. Is it integrated into say, Facebook photos? no.
  - Moldoteck 4y ago
    idk about llm, but they had released multimodal wav2vec, some variations of bert, transcoder if we talk pre gpt3 release and a lot more stuff after. Is it because of gpt or because some general advancements in AI&GPU power, we can't say
tinyhouse 4y ago
The metaverse pivot was a bold move that aligned well with their moto of connecting people. I always thought that the concept of the metaverse is silly but did think (and still do) that the potential of VR is huge if Meta can pull it off. For example, watching a live show or game in VR from your living room. (e.g., first row in a NBA game) The technology is not there yet. The Quest is nice but gets boring pretty quickly. They still don't have a killer app. That doesn't mean they won't find one in the next few years. As we saw with ChatGPT, many years of work can turn to an overnight success with one killer app.
- endisneigh 4y ago
  we'll have to agree to disagree. you don't need VR to "connect" with people. it's literally isolating. it's more or less a failure, hence the pivot towards AI. per the article many employees working on it are skeptical. if they are skeptical, it's basically already dead.
  - tinyhouse 4y ago
    I'm fine to agree to disagree but one comment about VR isolating people. It is today because the tech isn't there yet and most people use it to play a game by themselves. Think about an app where you can watch a live game with a friend who is 5000 miles away, feeling like you're together at the stadium. That's the opposite of isolating.
    - endisneigh 4y ago
      > Think about an app where you can watch a live game with a friend who is 5000 miles away, feeling like you're together at the stadium. That's the opposite of isolating. i suppose - or you can watched live stream together over video chat, and not have something over your face. you can do that today and not have to pay hundreds of dollars for hardware too :D
      - fatnoah 4y ago
        > i suppose - or you can watched live stream together over video chat, and not have something over your face. you can do that today and not have to pay hundreds of dollars for hardware too :D Speaking as someone who does sit with something on their face to "hang out" with friends thousands of miles away, it's an infinitely more immersive experience than video chat. We actually started with the chat route, but eventually gave getting together on the Quest a try, and it won us over immediately.

ripvanwinkle 4y ago

Part of the Meta's problem is this spiel about connecting people. Conversations internally are all about monetizing users and not losing any revenue due to new privacy regulations. Apart from the Metaverse which is also cloaked in "connecting people" there is no effort to take the technology and capabilities in the company and conquer new domains. Like Google getting into Office collaboration and Cloud computing etc. So that spiel about "connecting people" while being a bunch of whitewash is also restricting possibilities for the company

lostmsu 4y ago

> Zuckerboard went all in on AI Hindsight is 20/20

suzzer99 4y ago

The problem is I don't need any of this stuff and I know a ton of people who feel the same way. If FB's timeline had a chronological option like twitter, I would switch to it tomorrow, then mute people who bugged me. Every algorithm seems to end up optimizing for engagement, which basically means car wrecks, conspiracy theories, and other outrageous controversial stuff. Youtube will never stop recommending Jordan Peterson, no matter how many times you vote it down, because the algorithm has figured out that if it can get you into JP, it will open up a whole new world of engagement as you head further and further down the alt-right rabbit hole. So it's worth a periodic try at the hail mary, even for users who previously showed active aversion to JP.

pavlov 4y ago

Have you heard of LLaMa? Segment Anything? Those are just two major Meta AI projects released within the past two months. Generative AI is a major part of Zuck’s metaverse hopes and investment priorities.

PLenz 4y ago
To be fair, releasing the interesting parts of Llama was a decision made for instead of by them
endisneigh 4y ago
uh, what I'm saying is that all of that could've begun in 2014, not just now. clearly they're playing catch-up. Before OpenAI what language models did they release, exactly? how is segment anything playing into their metaverse strategy exactly? it's very clear that they're relasing this stuff to make it known that they're not going to fall behind to Microsoft (OpenAI) and Google. it clearly was not part of their original strategy around the metaverse, which is why it's all coming out now.
- ianbutler 4y ago
  https://ai.facebook.com/blog/roberta-an-optimized-method-for-pretraining-self-supervised-nlp-systems/ https://ai.facebook.com/blog/roberta-an-optimized-method-for... As just one example, which was (is?) one of the most used flavors of transformers from the BERT era.
  - endisneigh 4y ago
    I'm not talking about papers, I'm talking about a product offering, ala ChatGPT. Even Segment Anything isn't a real product. It's a glorified tech demo. Is it integrated into say, Facebook photos? no.
- Moldoteck 4y ago
  idk about llm, but they had released multimodal wav2vec, some variations of bert, transcoder if we talk pre gpt3 release and a lot more stuff after. Is it because of gpt or because some general advancements in AI&GPU power, we can't say

tinyhouse 4y ago

The metaverse pivot was a bold move that aligned well with their moto of connecting people. I always thought that the concept of the metaverse is silly but did think (and still do) that the potential of VR is huge if Meta can pull it off. For example, watching a live show or game in VR from your living room. (e.g., first row in a NBA game) The technology is not there yet. The Quest is nice but gets boring pretty quickly. They still don't have a killer app. That doesn't mean they won't find one in the next few years. As we saw with ChatGPT, many years of work can turn to an overnight success with one killer app.

endisneigh 4y ago
we'll have to agree to disagree. you don't need VR to "connect" with people. it's literally isolating. it's more or less a failure, hence the pivot towards AI. per the article many employees working on it are skeptical. if they are skeptical, it's basically already dead.
- tinyhouse 4y ago
  I'm fine to agree to disagree but one comment about VR isolating people. It is today because the tech isn't there yet and most people use it to play a game by themselves. Think about an app where you can watch a live game with a friend who is 5000 miles away, feeling like you're together at the stadium. That's the opposite of isolating.
  - endisneigh 4y ago
    > Think about an app where you can watch a live game with a friend who is 5000 miles away, feeling like you're together at the stadium. That's the opposite of isolating. i suppose - or you can watched live stream together over video chat, and not have something over your face. you can do that today and not have to pay hundreds of dollars for hardware too :D
    - fatnoah 4y ago
      > i suppose - or you can watched live stream together over video chat, and not have something over your face. you can do that today and not have to pay hundreds of dollars for hardware too :D Speaking as someone who does sit with something on their face to "hang out" with friends thousands of miles away, it's an infinitely more immersive experience than video chat. We actually started with the chat route, but eventually gave getting together on the Quest a try, and it won us over immediately.