Google wants governments to form a 'global AI corps'
washingtonpost.com
washingtonpost.com
Let me start by saying that Google is very likely separately pushing for regulation to make AI less accessible, and much of its public posturing is also very likely designed to drive up public interest in AI -- so to be clear I'm not saying any of the following in order to create the impression that Google is being philanthropic or has good motives; I do think that Google is overall a bad actor in this space.
However, in this very specific example, their poorly named "AI corps" appear to be less about controlling the dissemination of AI and more about retraining workers impacted by AI, investing resources into countries who have less funds to contribute to building their own models, increasing immigration opportunities for researchers, and adopting AI within the government as a tool (which has its own questions, but... different conversation).
Again, I'm not saying Google is a good actor. Google should not be trusted on AI regulation. And (opinion me) Google is over-hyping the space to serve its own interests. However, this very specific individual proposal doesn't seem to have the tone or really any of the problems that are being brought up in these comments. Nothing immediately jumped out to me as a proposal to lock down Open Source models or to form some kind of world government. The title is a little unfortunate and doesn't reflect the contents of the paper.
Here's the only phrase in the whitepaper that directly mentions a "global AI corps".
> Establishing a global AI corps that would include training and support programs to give workers hands-on experience in applying AI to solve new tasks on the job.
Basically "world governments should get together and make a plan to encourage everyone to use AI."
Is it problematic for Google to be asking the government to encourage adoption of its technologies? Potentially, if like me you believe those technologies are somewhat over-hyped right now (which is not the same as saying they're useless). But it's not a request to take over the world or close down competition. Those requests are going to come in separate papers. :)
—-
Above is what Chatgpt answered when i asked it about it.
• Governments and the private sector have a critical role to play in investing in R&D and AI infrastructure, including cloud infrastructure, compute capacity, and data, to ensure that researchers, technologists, and businesses have access to the tools needed to research, build and deploy AI. While countries have been focused on national-level approaches to this challenge, we should also think globally – and ensure that AI innovation is not the domain of only the richest countries and richest cities and regions. Countries should establish a Global Resource for AI Research (GRAIR). • In addition to investing in AI infrastructure, policymakers must also adopt strategies to enable researchers and innovators to convert ideas and data into new discoveries, products and services. This will require advancing pro-innovation legal frameworks that spur dynamism across the entire ecosystem. • Finally, because AI is by its nature a cross-border technology, individual policy efforts must be tethered to strong trade and investment policies that support trusted international collaboration on AI, including cross-border data flows essential to AI development and deployment. ... we should ensure that academic researchers in a broader range of countries globally, including in the developing world, have access to compute resources – to ensure that they can help advance AI innovation and promote AI research. For this reason, governments should come together to establish a Global Resource for AI Research (GRAIR) to ensure that we enable AI research globally. The GRAIR would serve as a shared AI research resource and clearinghouse to make compute, data, and software resources available to academics, start-ups and small and medium enterprises conducting AI development and research. Beyond providing technical resources, the GRAIR should also provide support for countries at different levels of development to build up domestic AI workforce capabilities, including application developers and researchers.
We've seen it happen before.
It's painful watching Google throw away opportunity after opportunity to real make what people want in their single-minded pursuit of more revenues from ads.
Unlike a company like Apple that has diversified their income streams from hardware and can easily branch into building their own search engine, Google's only priority is chuck ads wherever they fit. The only way they'll catch up with OpenAI's speed of innovation is if (and that's a big IF) sama & co. squanders their lead. Judging by the past 1 year (strategic partnerships with Microsoft, branching out into a platform model that disrupts GPT wrappers preemptively, faster, cheaper experiences, $1B ARR), I assume they won't.
I mean, OpenAI and Microsoft are doing it the other way around - "We should regulate these smaller companies/FOSS models so tehy can't compete with us." Altman literally begged the Congress to regulate them a few months ago.
I can't wait until the tech becomes accessible enough that we don't have to rely on these companies for most common tasks with LLMs and tell them to gently piss off.