OpenAI loses fourth company manager over ethical disagreements
“Designing machines that are smarter than humans is an inherently dangerous endeavor. OpenAI bears this immense responsibility on behalf of all humanity.”
Jan Leike
Former head of security at OpenAI, co-head of the unit responsible for ethics and human protection
The OpenAI tech ship is gradually losing its crew: Jan Leike, one of the company’s main security researchers, is the fourth member to resign this month.
He was the “co-responsible for the superalignment” at OpenAI, a position that involved ensuring that artificial intelligence (AI) systems “respect the values and goals of humanity”, writes the British daily The Guardian. He explains today that he no longer finds these famous values in the way the company operates.
This sudden departure takes place in the wake of that of Ilya Sutskever, co-founder of OpenAI and emblematic figure of the sector, and those of “two senior security officials at OpenAI who recently resigned” for ethical disagreements, underlines the Guardian.
“Over the years, the importance given to safety and compliance with procedures has taken a back seat in favor of flashy products,” writes Jan Leike on his X account, regretting that the company does not invest “more resources in security, social impacts, privacy and security” of his models.
Yesterday was my last day as head of alignment, superalignment lead, and executive @OpenAI.
— Jan Leike (@janleike) May 17, 2024
This resignation came on the eve of a global summit on AI organized in Seoul, the central theme of which is the surveillance of this technology. Since its opening, sixteen AI giants, including OpenAI, have committed to “do not develop software that could threaten humanity” by signing a charter, reports the London daily The Times.
“Extreme risks for humanity”
Well calculated timing or coincidence? Still, with this charter, designed for “allay fears about technology”, AI companies commit to “mitigate extreme risks to humanity”, one of the main ones of which is “the loss of control of a superintelligent system by humans”.
At the heart of these fears is “the belief that one day AI systems will be smarter than human experts”, raise it Times. A fear that Jan Leike, who resigned from OpenAI, strongly shares.
