Sources detail the pressure inside Meta and Google to move faster on AI due to the surge of attention around ChatGPT, potentially sweeping safety concerns aside
Google, Facebook and Microsoft helped build the scaffolding of AI. Smaller companies are taking it to the masses, forcing Big Tech to react.
Washington Post
Context & Ripple Effects
Google had already been weighing the reputational risk of releasing LaMDA against the danger of losing ground to ChatGPT, as described in Google's earlier hesitation over LaMDA. The attention around ChatGPT turns that internal product-timing dilemma into a broader competitive problem for Google and Meta.
Subsequent coverage describes Microsoft and Google accepting more risk relative to their AI ethics guidelines and Google limiting research sharing until work had been productized, linking the race to both governance and publication choices.
First-order effects
Meta and Google face immediate internal pressure to accelerate AI development and releases in response to ChatGPT's public momentum, with safety concerns at risk of being deprioritized.
Google's prior caution around LaMDA becomes harder to sustain when a competing chatbot has made consumer-facing AI a visible competitive category.
Second-order effects
Microsoft and Google are pushed toward a faster risk calculus, reflected in later reporting that ChatGPT's success made them more willing to depart from their established AI-ethics posture.
Google's move toward withholding research until products were ready shows the competition reaching upstream: research disclosure becomes part of product strategy rather than only scientific exchange.
Third-order effects
If major labs continue treating public AI releases as competitive deadlines, internal safety review and research openness will be governed increasingly by product-race incentives rather than each company's stated principles.
The pattern points to AI industrialization in which control of the route from research to consumer product becomes a strategic advantage for the largest platforms.
The trend: Consumer adoption of ChatGPT is turning AI development from a research and reputation question into a speed-to-market contest among major platforms.
LLMs have no physical intuition as they are trained exclusively on text. They may correctly answer physical intuition questions *if* they can retrieve answers to similar questions from their vast associative memory. But they may get the answer *completely* wrong 1/ https://twitte…
There are many frustrating bits in this. This one I take personally. While I can only speak to Cloud's #ResponsibleAI work under my leadership (+ I'm confident, the current team) we NEVER made decisions b/c of PR. We made decisions based on risk of HARMS. https://www.washingtonpo…
How mighty companies get disrupted, in real time. Great @washingtonpost on the obstacles to Google and Meta succeeding in AI https://www.washingtonpost.com/ ... https://twitter.com/...
@ylecun The problem I keep thinking about is that ‘make things up’ means different things in different domains. 90% accurate can mean ‘almost perfect’ or ‘totally wrong’ even for different kinds of text, and images even more so.
Excellent WaPo article about large language models and chatbots that corroborates what I've been posting recently: they are useful but they make stuff up. They detail the reasons why large tech cos have been hesitant to release such things for public use. https://www.washingtonpo…
I think we can all agree that what's been holding us back lately is tech giants being being cautious and beholden to ethicists. Glad we're getting more headlong charges into murky waters without regard for risk, now. I'm sure it will be fine. https://twitter.com/...
For years, tech giants like Google and Facebook have been developing the underlying tech behind tools like ChatGPT and Stable Diffusion—but had been cautious about releasing them due to concerns over safety, bias, and misinfo. That's changing now. https://www.washingtonpost.com/ …
More generative AI innovations. Watch as Meta and Google roll out everything they've been working on over the next 6 weeks, as a response to ChatGPT. It's going to be wild. https://twitter.com/...
🚨 Correcting @ylecun here: @Meta/@facebook *did* release a #ChatGPT-like thing, the highly-anticipated #Galactica, but it was withdrawn just three days after its release, following a deluge of trolling and accusations of bias! The internet remembers: https://techunwrapped.com/...…
@ylecun ... Ask Meta to spin-off a semi-independent section of FAIR and release under a non-profit setup with exclusive deal to Meta (profit) That's how Microsoft is working around it. Maybe name it OpenAIR 😉
@DataChaz ... You are making my point. Ask yourself why Galactica was pilloried and crucified while chatGPT (which has similar flaws) was welcome as the second coming of the Messiah?
I came up with the example above during my interview with Alex Kantrowitz for his podcast. ChatGPT's answer sounds good, but it's exactly backwards. Funny thing is that Alex read the response quickly to me during the interview and didn't immediately realize it was wrong! 2/
@powerbottomdad1 Yeah that meta chatbot sucks. It's going down a different route than how chatGPT is built though. But lambda is kinda good, wish that big G would release it
yann lecun yesterday: open ai is simply just not ahead of google and meta. they can easily make a product like chatgpt a conversation with metas new chatbot today: [Screenshots of answers from Meta's chatbot]
@parismarx I don't know if it was in your podcast or mentioned by someone else, but the ChatGPT and recent resurgence of desperate AI hype cycle feels like it stems from L after L after L for the “next big thing in tech”. They no longer want a win, they NEED a win.
@parismarx Foretold by Sam Harris in one of his Ted Talks, he noted that, once we saw successful examples of AI, it would become an arms race. And in the race, big mistakes will be made. (He was talking about countries, similar to the space race, but theory applies to tech compan…
@parismarx It's amazing how freaked-out tech companies get over demos that (1) don't make any money and (2) are jury-rigged or conceptually limited in some way that makes them totally unmarketable.
Race to the bottom indeed. I think this is good reporting, though it seems to be very eager to take the perspective of engineers who are “frustrated” with their employers for being cautious. https://www.washingtonpost.com/ ... #AIhype #MathyMath #ChatGPT
MusicLM: Generating Music From Text (sound on 📣) project page: https://google-research.github.io/ ... arXiv: https://arxiv.org/... https://twitter.com/...
Educated skeptics are so important. Yann highlights a limitation of current LLMs and tools like ChatGPT. Some might consider this as being a hater, but I consider this as the only way to understand how this stuff actually works. https://twitter.com/...