Sekėjai

Ieškoti šiame dienoraštyje

2023 m. lapkričio 21 d., antradienis

 

Sam Altman played with death of humanity

"Corporate power struggles aren't rare, and most companies survive them. But the OpenAI board's sudden ouster of CEO Sam Altman last week looks increasingly like a misguided kamikaze run. It's hard to see who benefits from the melodrama besides opponents of innovation.

News of Mr. Altman's removal late Friday shocked Silicon Valley. Mr. Altman helped launch OpenAI in 2015 with the mission of responsibly developing artificial intelligence to benefit humanity. He has since become the technology's public face, with media appearances and testimony to Congress.

The reasons for Mr. Altman's defenestration are murky. The board without elaboration said he hadn't been consistently candid in his communications, and the company's chief operating officer in a memo to employees cited a "breakdown in communications." But recent reports suggest the coup was the result of a dispute over the future of AI.

OpenAI was initially organized as a nonprofit with donations from Silicon Valley leaders. But to rapidly advance its large-language-model capabilities, it needed to raise more money. The result was an unorthodox governance structure in which a nonprofit with an altruistic mission and board of directors controls a for-profit subsidiary whose profits are capped.

This halfway house let OpenAI raise billions of dollars in private capital from the likes of Sequoia Capital, Andreessen Horowitz and Microsoft while preserving its public-spirited goals. A for-profit can also use stock options to recruit talent.

Mr. Altman collaborated with Microsoft to raise some $13 billion in capital and run its programs on the tech titan's servers. Microsoft obtained a reported 49% stake in OpenAI and can provide its cloud-computing customers access to OpenAI's advanced models -- a large reason Microsoft's market valuation has soared above $2.8 trillion this year.

By almost any measure, Mr. Altman was a successful CEO. Two million developers, including more than 92% of Fortune 500 companies, use OpenAI, and ChatGPT boasts more than 100 million weekly active users. But use of its large-language models is growing so fast that it is struggling to find the computing power to keep up.

At the same time some OpenAI board members fret about OpenAI's growth. They worry that its rapidly advancing capabilities could undermine safety and the company's altruistic mission. Two board members have ties to the so-called effective altruism movement, which fears that AI could present an existential threat to humanity.

Some board members resisted Mr. Altman's push to raise capital to expand OpenAI's capabilities. Mr. Altman reportedly wanted to democratize ChatGPT by letting users create their own chatbots to perform specialized tasks. The board's high-minded altruists worry this could cause AI to grow out of control.

But raising more capital is essential to managing the technology's risks. Safety and growth aren't incompatible, though AI skeptics treat them as such. All of this is important context for Mr. Altman's sacking.

Investors say they might have to write down their stakes, which could make it harder for OpenAI to raise more capital. The board's apparent failure to consult investors before removing Mr. Altman doesn't build confidence in its oversight. The corporate good-governance crowd hails "independent" boards, but OpenAI shows this isn't always a virtue.

Microsoft has offered Mr. Altman a job, though this could be a negotiating tactic to get the board to bring him back. More than 700 of OpenAI's 770 employees have threatened to quit unless the board resigns and reinstates Mr. Altman, which underscores how much goodwill he had at the company. What board seeks to destroy its own company?" [1]
1.

 1. Ineffective Altruists vs. Sam Altman. Wall Street Journal, Eastern edition; New York, N.Y.. 21 Nov 2023: A.14.


 

Komentarų nėra: