Of all the think pieces, manifestos, hand wringing, and news analysis I’ve read about AI threats and opportunities, the one piece that feels critical to consider is a plea from Mustafa Suleyman, CEO of Microsoft AI. Its subhead lays out its core argument: “AIs do not have rights, feelings, or consciousness. And we must not train them to act as though they do.”
Suleyman opens with a very clear point of view. Let’s call it Suleyman’s Conjecture:
“AIs are not conscious. They do not feel, experience, or suffer. They do not have innate preferences or underlying motivations. They are sequence completion engines, internally hollow, designed to follow instructions, and accomplish goals set by humans.
If humanity is to flourish in the 21st century, that is how they must remain.”
I found myself nodding along as I read those words, well aware that many in the Valley – and in academia – strongly disagree. If AI is truly “intelligent,” they argue, then we must treat AI as a new kind of life form, one worthy of rights and privileges. To do anything less risks alienating AI into a confrontation we’d surely lose.
I understand this line of reasoning, but I find myself repelled by it. And I sense that when the histories of this age are written – if we make it to that future state – Suleyman’s conjecture will prove a pivotal chapter. Well-intentioned, intelligent people will find themselves divided by the issue – just as they do on fundamental questions of human welfare like abortion and the death penalty.
However, much of Suleyman’s essay comprises a (rather elegant) tear-down of Anthropic, Microsoft’s main competitor in the race to build ever-powerful AI. While it’s true that Anthropic has intentionally anthropomorphized its platform, Suleyman’s admitted conflict of interest colors his argument as motivated by competitive positioning and profit-seeking. But he is correct when he states “This issue needs urgent public debate.” I agree.
One place to start is Microsoft’s recently published “Humanist AI Code of Conduct,” which states Suleyman’s conjecture unequivocably: “The purpose of technology is to serve humanity, and any technology that cannot remain within human control should be rejected.” The document is very long – as is Anthropic’s Claude Constitution – but if you are serious about understanding this debate, reading both is a very good place to start.*
—
*Another is to follow Reid Hoffman’s thinking – including his recently published “AI for All Americans,” from which the image above is taken. Also worth thinking about: “AI As Normal Technology – an alternative to the vision of AI as a potential superintelligence.”