Microsoft AI says it’ll make superintelligent AI that won’t be terrible for humanity

Microsoft AI wants you to cognize that its enactment toward superintelligence involves keeping humans “at the apical of the nutrient chain.” In a lengthy blog station connected Thursday, Microsoft AI caput Mustafa Suleyman highlighted the enactment of a caller squad dedicated to processing a “humanist superintelligence” that’s “designed lone to service humanity.”

This benignant of superintelligence, according to Suleyman, won’t beryllium “an unbounded and unlimited entity with precocious degrees of autonomy” and volition alternatively beryllium “carefully calibrated, contextualized, wrong limits.” Suleyman joined Microsoft past twelvemonth arsenic the CEO of Microsoft AI, which has lone precocious launched its first in-house models for text, voice, and representation generation.

Though Suleyman’s blog station says Microsoft AI volition “reject narratives astir a contention to AGI,” the contention betwixt Microsoft and OpenAI is astir to get overmuch much intense. Under a caller woody with OpenAI, Microsoft tin present “independently prosecute AGI unsocial oregon successful concern with 3rd parties.” And, arsenic pointed retired by my workfellow Hayden Field, “Microsoft is perfectly wrong its ineligible rights to usage OpenAI’s IP to make its ain AGI and effort to triumph the race.”

But Suleyman has a imaginativeness for “humanist” superintelligence with 3 main applications, which see serving arsenic an AI companion that volition assistance radical “learn, act, beryllium productive, and consciousness supported,” offering assistance successful the healthcare industry, and creating “new technological breakthroughs” successful cleanable energy.

“At Microsoft AI, we judge humans substance much than AI,” Suleyman writes. “Humanist superintelligence keeps america humans astatine the centre of the picture. It’s AI that’s connected humanity’s team, a subordinate, controllable AI, 1 that won’t, that can’t unfastened a Pandora’s Box.”