5 ms·

Mark Zuckerberg Shakes Up Meta's A.I. Efforts, Again

59 points by voxadam 1y ago

voxadam 1y ago
https://archive.is/lTWs7 https://archive.is/lTWs7
- icarusfowl 1y ago
  https://www.nytimes.com/2025/08/19/technology/mark-zuckerberg-meta-ai.html?unlocked_article_code=1.fU8.WaFO.eCCxlNf5qg-L&smid=url-share https://www.nytimes.com/2025/08/19/technology/mark-zuckerber...
kilroy123 1y ago
I think no matter what Mark does to shake things up. ~Facebook~ Meta will always be out-innovated by a startup and forced to keep acquiring and gobbling up anything that becomes remotely successful.
- nine_zeros 1y ago
  > I think no matter what Mark does to shake things up. ~Facebook~ Meta will always be out-innovated by a startup This happens to all companies that end up with - pardon my French - a shithole culture.
  - lazyeye 1y ago
    They "trust me"...Dumb f*ks
    - Nevermark 1y ago
      I would disagree if this were a spurious quote that wasn't highly indicative of the preponderance of his record. All the ethical quandaries where Facebook has been caught red handed. Over and over. Habitual denial of obvious problems. The gross ineptitude and irresponsibility that facilitated actual genocidal coordination, in languages with no Facebook monitors. The consistently slippery talk we get from him via PR and Senate interviews. And the deeply ethically challenged core business model of surveillance by any means not deemed illegal, and then some, as a means of individualizing the gaming of people's psychology - without transparent disclosure or a required opt-in.
      - lazyeye 1y ago
        Agreed. They were also happy to give up a US citizens to the CCP in the hope of gaining marketshare in China. There was a senior whistleblower in Facebook that revealed this a little while ago. https://www.youtube.com/watch?v=f3DAnORfgB8 https://www.youtube.com/watch?v=f3DAnORfgB8
moralestapia 1y ago
EOL for LeCunn.
- redleader55 1y ago
  I follow him on LinkedIn: he works on the things that will come in 10 years, not on the products that Meta ships in 1-2 years. LeCunn is going to be there for some time.
  - rlanday 1y ago
    Yann LeCun is 65. It is extremely unlikely his future research contributions will amount to anything close to what he has already done.
- HarHarVeryFunny 1y ago
  No - he's head of FAIR which is still there. The new "reorg" isn't really much of a reorg, with the four AI groups now being: TBD (LLMs / "super-intelligence") FAIR (LeCun - JEPA and other long-term research) Productization Infrastructure
ml-anon 1y ago
MSL can already be written off as a failure. Zuck didn’t get the people he wanted despite literally offering billions. He then attached MSL to his other failed GenAI org which was already an astounding failure of leadership from the now triple headed team (wang, nat and zuck) which resulted in people fighting for budget, scope and prestige. Meanwhile you have 1000x comp inequality among ICs and in some cases people getting 10x higher offers than others just because they happened to finish interviews a week later. These people couldn’t organize a piss up in a brewery, good luck with AGI.
afavour 1y ago
I really wonder how Zuckerberg will be looked back on. It's seemed obvious that he's an incredible CEO given the growth we saw with Facebook, but maybe he was just in the right place at the right time and had a cash cow that was difficult to mess up? Because when I look at the Metaverse and I look at this AI stuff... I dunno. Feels like throwing spaghetti at the wall.
- galoisscobi 1y ago
  I think he’ll be looked back as one hit wonder, lacking any ethics to grow his company. He pumped a whole bunch of money in VR that didn’t seem to go anywhere and now AI seems like a fine direction to go in, copying the herd, with dangerous disregard to ethics.
  - atonse 1y ago
    Agree with you about ethics but it was a master stroke for him to buy both Instagram and WhatsApp.
    - herbturbo 1y ago
      He was just buying users of popular products in lieu of attracting new ones to his own. Meta has subsequently ruined IG and seems busy working up new ways to ruin WhatsApp.
      - atonse 1y ago
        Sure but seeing where his users were going and buying those companies at what are now very low prices (especially instagram) were a big deal. Not so much with Oculus though. The whole metaverse thing ended up being very cringe.
  - nemothekid 1y ago
    >I think he’ll be looked back as one hit wonder, lacking any ethics to grow his company. I can't take this seriously - it reeks of hindsight bias. Zuckerberg's "one hit" was thefacebook.com. After facebook.com went public he, seemingly immediately, decided to buy Instagram and go all in on mobile. At the time, people thought he had way too much power due to the stock structure and many people thought his bet on mobile would come crashing down. We can see, in hindsight, that it was a prescient move - one that many others missed, or were late to. (e.g. Google and Microsoft). I don't think there is a single founder/CEO in the 21st century that is performing better than Zuckerberg. I understand he's not a likable guy, and neither are are his products. The only facebook product I use is arguably React - I've deactivated my facebook long ago, and I no longer have instagram. I don't even have Whatsapp. But if you look at the metrics they aren't deniable. Facebook figured out how to print money in social media while every other social media company struggles to have a quarter of the profitability. A lot of people point to "he just bought Instagram", without seriously interrogating the fact that many apps have been bought and squandered.
11101010001100 1y ago
ok, but I want to know if these superstars still had to leetcode interviews.
- gedy 1y ago
  Yeah I mean I know these are experienced AI experts, but they could be faking or coasting. We better give a twentysomething Sr. Software Engineer veto power over these hires just to be sure./s
  - Der_Einzige 1y ago
    Unironically a lot of AI researchers are trash at leetcode and will cheat on interviews.
    - BobaFloutist 1y ago
      How good can they be at AI if they can't even cheat effectively in interviews
emocin 1y ago
[flagged]
bgwalter 1y ago
$72 billion per year to create yet another chatbot. Imagine if this were spent on useful research or housing.
- nomel 1y ago
  I don't think that's fair. It's R&D for a potential infinite money/effort hack. Unless you're in the camp that thinks only meat can be intelligent, depending on your perspective, it can be considered somewhere between useful research and critical research, including for the pursuit of things like "housing". Maybe all the learnings of the specific architectures/implementations will be thrown out in the end, but that's not some rare occurrence with R&D/progression. If you think their goal is to make chatbots, then maybe this won't make sense to you.
  - dmonitor 1y ago
    So what's the goal? Summon a god?
    - Balgair 1y ago
      There's this thing about lovecraftian ur gods. You have to have these people that can speak these horrible languages, study for years unimpeded for likely no goal, can finance expeditions to the bottom of the ocean and fail again and again, sacrifice the right victims at the right times, etc. These people that will awaken Cthulhu have to be some of the most brilliant in human history (so says the lovecraftian mythos), and then they have to throw away all their brains and money and time to awaken the 'do not awaken under any circumstances' dark god of terror and soul consuming ambivalence. Like, there isn't anyone out there that is that stupid. Totally breaks the immersion of the worldbuilding, right? ...right?
    - nomel 1y ago
      Do you think you're a god? You're are a general intelligence, that has the capacity to learn and create. This is the stated goal of most of these companies: make a general intelligence that allows autonomous agents (possibly controlling robots) to do general, human like, work with sufficient, and ideally superior, competency. No silly claims of gods required. It's interesting how these conversations usually turn irrational, immediately.
darepublic 1y ago
> The new team has discussed making Meta’s next A.I. model “closed,” which would be a major departure from the company’s longtime philosophy of “open sourcing” its models. A closed model keeps its underlying code secret, while an open-source A.I. model can be built upon by other developers. Gotta assume this was the plan all along. And would have been plan A if they weren't caught flat footed by chatgpt

59 points by voxadam 1y ago

voxadam 1y ago

https://archive.is/lTWs7 https://archive.is/lTWs7

icarusfowl 1y ago
https://www.nytimes.com/2025/08/19/technology/mark-zuckerberg-meta-ai.html?unlocked_article_code=1.fU8.WaFO.eCCxlNf5qg-L&smid=url-share https://www.nytimes.com/2025/08/19/technology/mark-zuckerber...

kilroy123 1y ago

I think no matter what Mark does to shake things up. ~Facebook~ Meta will always be out-innovated by a startup and forced to keep acquiring and gobbling up anything that becomes remotely successful.

nine_zeros 1y ago
> I think no matter what Mark does to shake things up. ~Facebook~ Meta will always be out-innovated by a startup This happens to all companies that end up with - pardon my French - a shithole culture.
- lazyeye 1y ago
  They "trust me"...Dumb f*ks
  - Nevermark 1y ago
    I would disagree if this were a spurious quote that wasn't highly indicative of the preponderance of his record. All the ethical quandaries where Facebook has been caught red handed. Over and over. Habitual denial of obvious problems. The gross ineptitude and irresponsibility that facilitated actual genocidal coordination, in languages with no Facebook monitors. The consistently slippery talk we get from him via PR and Senate interviews. And the deeply ethically challenged core business model of surveillance by any means not deemed illegal, and then some, as a means of individualizing the gaming of people's psychology - without transparent disclosure or a required opt-in.
    - lazyeye 1y ago
      Agreed. They were also happy to give up a US citizens to the CCP in the hope of gaining marketshare in China. There was a senior whistleblower in Facebook that revealed this a little while ago. https://www.youtube.com/watch?v=f3DAnORfgB8 https://www.youtube.com/watch?v=f3DAnORfgB8

moralestapia 1y ago

EOL for LeCunn.

redleader55 1y ago
I follow him on LinkedIn: he works on the things that will come in 10 years, not on the products that Meta ships in 1-2 years. LeCunn is going to be there for some time.
- rlanday 1y ago
  Yann LeCun is 65. It is extremely unlikely his future research contributions will amount to anything close to what he has already done.
HarHarVeryFunny 1y ago
No - he's head of FAIR which is still there. The new "reorg" isn't really much of a reorg, with the four AI groups now being: TBD (LLMs / "super-intelligence") FAIR (LeCun - JEPA and other long-term research) Productization Infrastructure

ml-anon 1y ago

MSL can already be written off as a failure. Zuck didn’t get the people he wanted despite literally offering billions. He then attached MSL to his other failed GenAI org which was already an astounding failure of leadership from the now triple headed team (wang, nat and zuck) which resulted in people fighting for budget, scope and prestige. Meanwhile you have 1000x comp inequality among ICs and in some cases people getting 10x higher offers than others just because they happened to finish interviews a week later. These people couldn’t organize a piss up in a brewery, good luck with AGI.

afavour 1y ago

I really wonder how Zuckerberg will be looked back on. It's seemed obvious that he's an incredible CEO given the growth we saw with Facebook, but maybe he was just in the right place at the right time and had a cash cow that was difficult to mess up? Because when I look at the Metaverse and I look at this AI stuff... I dunno. Feels like throwing spaghetti at the wall.

galoisscobi 1y ago
I think he’ll be looked back as one hit wonder, lacking any ethics to grow his company. He pumped a whole bunch of money in VR that didn’t seem to go anywhere and now AI seems like a fine direction to go in, copying the herd, with dangerous disregard to ethics.
- atonse 1y ago
  Agree with you about ethics but it was a master stroke for him to buy both Instagram and WhatsApp.
  - herbturbo 1y ago
    He was just buying users of popular products in lieu of attracting new ones to his own. Meta has subsequently ruined IG and seems busy working up new ways to ruin WhatsApp.
    - atonse 1y ago
      Sure but seeing where his users were going and buying those companies at what are now very low prices (especially instagram) were a big deal. Not so much with Oculus though. The whole metaverse thing ended up being very cringe.
- nemothekid 1y ago
  >I think he’ll be looked back as one hit wonder, lacking any ethics to grow his company. I can't take this seriously - it reeks of hindsight bias. Zuckerberg's "one hit" was thefacebook.com. After facebook.com went public he, seemingly immediately, decided to buy Instagram and go all in on mobile. At the time, people thought he had way too much power due to the stock structure and many people thought his bet on mobile would come crashing down. We can see, in hindsight, that it was a prescient move - one that many others missed, or were late to. (e.g. Google and Microsoft). I don't think there is a single founder/CEO in the 21st century that is performing better than Zuckerberg. I understand he's not a likable guy, and neither are are his products. The only facebook product I use is arguably React - I've deactivated my facebook long ago, and I no longer have instagram. I don't even have Whatsapp. But if you look at the metrics they aren't deniable. Facebook figured out how to print money in social media while every other social media company struggles to have a quarter of the profitability. A lot of people point to "he just bought Instagram", without seriously interrogating the fact that many apps have been bought and squandered.

11101010001100 1y ago

ok, but I want to know if these superstars still had to leetcode interviews.

gedy 1y ago
Yeah I mean I know these are experienced AI experts, but they could be faking or coasting. We better give a twentysomething Sr. Software Engineer veto power over these hires just to be sure./s
- Der_Einzige 1y ago
  Unironically a lot of AI researchers are trash at leetcode and will cheat on interviews.
  - BobaFloutist 1y ago
    How good can they be at AI if they can't even cheat effectively in interviews

emocin 1y ago

[flagged]

bgwalter 1y ago

$72 billion per year to create yet another chatbot. Imagine if this were spent on useful research or housing.

nomel 1y ago
I don't think that's fair. It's R&D for a potential infinite money/effort hack. Unless you're in the camp that thinks only meat can be intelligent, depending on your perspective, it can be considered somewhere between useful research and critical research, including for the pursuit of things like "housing". Maybe all the learnings of the specific architectures/implementations will be thrown out in the end, but that's not some rare occurrence with R&D/progression. If you think their goal is to make chatbots, then maybe this won't make sense to you.
- dmonitor 1y ago
  So what's the goal? Summon a god?
  - Balgair 1y ago
    There's this thing about lovecraftian ur gods. You have to have these people that can speak these horrible languages, study for years unimpeded for likely no goal, can finance expeditions to the bottom of the ocean and fail again and again, sacrifice the right victims at the right times, etc. These people that will awaken Cthulhu have to be some of the most brilliant in human history (so says the lovecraftian mythos), and then they have to throw away all their brains and money and time to awaken the 'do not awaken under any circumstances' dark god of terror and soul consuming ambivalence. Like, there isn't anyone out there that is that stupid. Totally breaks the immersion of the worldbuilding, right? ...right?
  - nomel 1y ago
    Do you think you're a god? You're are a general intelligence, that has the capacity to learn and create. This is the stated goal of most of these companies: make a general intelligence that allows autonomous agents (possibly controlling robots) to do general, human like, work with sufficient, and ideally superior, competency. No silly claims of gods required. It's interesting how these conversations usually turn irrational, immediately.

darepublic 1y ago

> The new team has discussed making Meta’s next A.I. model “closed,” which would be a major departure from the company’s longtime philosophy of “open sourcing” its models. A closed model keeps its underlying code secret, while an open-source A.I. model can be built upon by other developers. Gotta assume this was the plan all along. And would have been plan A if they weren't caught flat footed by chatgpt