In an internal memo, Microsoft CEO Satya Nadella weighed in on the growing debate over whether the AI industry is moving too fast.
Nadella said AI should remain under human control, endorsed broader third-party testing, and warned that companies must take time to make the technology safety, according to the memo, which was posted to a company message board late Monday.
“As the stakes get higher, one should take all the time they need,” Nadella wrote. “If not you will anyway lose permission to operate.”
His comments follow dire warnings from some AI researchers, who said last week that the technology might kill all humans.
This weekend, Anthropic CEO Dario Amodei proposed slower AI development and said his company would allow outside monitoring of its model progress. OpenAI CEO Sam Altman concurred on third-party monitoring and expressed support for slowing deployment when necessary to ensure alignment and safety.
Microsoft AI on Monday released a draft of its 37-page code of conduct for its “humanist” AI models. “Things we have worried about for a long time in theory have become very real,” Microsoft AI CEO Mustafa Suleyman wrote.
Read the full memo
“There was a lot of conversation about AI over the weekend. As a platform company, for us the focus remains on putting our approach into practice across three areas.
First, we need to be clear about the design goal. As I shared, AI has to remain under human control and broadly benefit society. That means building a frontier ecosystem where both closed and open-source models can thrive, and being deliberate about pacing when it comes to alignment. I think of this like testing. We all do deliberate pacing when it comes to pre-release R&D and testing. And having more outside/3P testing/testers is a good thing. And as the stakes get higher, one should take all the time they need! If not you will anyway lose permission to operate. That is how we work every day. We will approach this with all the work we are doing to ensure observability at every layer of stack and use to enforce governance and containment. We will also use this as the guide to how we build our models. Mustafa published the “Code of Conduct” that underlies our own first party MAI models today for public consultation.
Second, for our customers, the key is making sure they stay in control of their unique and tacit knowledge, so they can build their own continuous learning loop/hill climbing machine, without becoming dependent on any one model provider. This is really what Foundry enables, and there was a great demo showing how it makes long-running agents enterprise ready, with security, safety guardrails, FinOps, etc. all built in from the start.
And third, we have to earn the permission to build the infrastructure that powers all of this by ensuring the benefits are felt locally. Our datacenter in Quincy is probably the best example of what it looks like when we grow alongside a community over time, and Brad shared that story on Sunday.
So really, it’s one approach across three layers: how we build at the frontier, how we enable our customers to stay in control, and how we make sure the economic benefits diffuse broadly across communities and countries.
And importantly, as a company we can point to real action behind the words.”
Have a tip? Contact this reporter via email at [email protected] or Signal at +1-425-344-8242. Use a personal email address and a nonwork device; here’s our guide to sharing information securely.
Read the original article on Business Insider
The post Microsoft CEO Satya Nadella posts an internal warning to employees about AI safety appeared first on Business Insider.
