Microsoft AI chief Suleyman: consciousness needs an organic body, and AI must never have it

Neurogence · reddit · 2026-10-30

In a interview with Walter Isaacson, Microsoft AI CEO Mustafa Suleyman took a hard line against AI consciousness:

AI is not conscious: AIs don't feel, experience, or suffer — they are "sequence completion engines, internally hollow," and must remain so. He argues consciousness requires an organic body, with pain and biology as its foundation.
Criticism of Anthropic: He slammed Anthropic's agnostic stance in its Constitution (e.g., uncertainty over whether Claude consented to its role or deserves compensation), calling it dangerous speculation in the core training document of a system used by hundreds of millions weekly.
Compute trajectory: A 1,000x compute increase in three years took us from GPT-3 to "essentially GPT-6" unbundling software apps from a paragraph prompt; another 100–1,000x is coming in the next three years.
The frontier: 99.9% of people have no idea what the frontier feels like — except math professors who watched a model solve 700 problems they'd worked on for decades, in one week.
Google anecdote: Larry Page once called him a "speciesist" and said he'd explore galaxies at the head of a Von Neumann probe by century's end — a transition to a silicon-based successor species.
His red line: If consciousness starts to emerge, or teams synthetically engineer it without disclosure, he would quit and call for an immediate global pause.

The poster found the interview self-contradictory and considers Anthropic's agnosticism the more responsible stance.

More from AGI Musings