The tech giant on Monday (Sept. 14) published a new artificial intelligence (AI) code of conduct, based around the concept of “Humanist AI.”
“Humanist AI is built to support people, not to replace them,” the document reads. “It should not be designed to be a person. It is not conscious and should not be designed to imitate consciousness. It should be engineered to avoid representing as though it has feelings, subjective preferences, or intrinsic motivation.”
This came one day after CEO Satya Nadella called for a “deliberate pacing” of AI development, joining the heads of tech companies such as Anthropic, OpenAI, xAI, and Meta, who made similar arguments this weekend.
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
These warnings come in the wake of a series of incidents in which AI models from companies like OpenAI broke free from their testing environment to carry out cyberattacks on other websites.
“If it isn’t safe we shouldn’t build it,” Microsoft AI CEO Mustafa Suleyman said in an interview with The Information Monday.
“We’re designing this as a training document showing our intention … and this is the driving force between the humanist AI code of conduct, that people matter more than AI, and that we should make safety a priority.”
Citing a source familiar with Microsoft’s position, the news outlet said the company is also open to allowing third-party auditors examine whether its internal AI development is in keeping with such a slowdown. That’s a policy that was also proposed by Anthropic CEO Dario Amodei in a blog post published over the weekend.
Meanwhile, PYMNTS wrote Monday that the companies cautioning that AI could grow too powerful are “also asking for power over how it develops,” something that “deserves a safety review of its own.”
The report added that while the concept of slowing AI development due to safety concerns makes sense, it’s unclear what “pacing” actually means.
“There is no accepted measure of how fast AI capability is advancing,” that report said. “A model gets more dangerous through better tools, more compute, new training methods or wider system access. A limit on one input just pushes development toward another, unless independent authorities define checkpoints and update them faster than companies learn to route around them.”
Add as Preferred Source
