To the Editor:
Re “The A.I. Safety Experts Can’t Sleep, and Neither Should You,” by Stephen Witt (Opinion guest essay, Sept. 13):
Mr. Witt is right to sound the alarm on the dangers of leaving the rapidly advancing A.I. industry largely to regulate itself. But despite the recent urging by Anthropic’s chief executive, Dario Amodei, to slow down A.I. development (front page, Sept. 13), the prospect seems unlikely.
At this month’s G20 Innovation Ministerial in Chapel Hill, N.C., hosted by the Department of Commerce and the White House Office of Science and Technology Policy, the United States urged other nations toward a hands-off approach to A.I. regulation. Elon Musk has recently criticized Europe’s more risk-based regulatory system because it “inhibits progress” and “slows it down.”
That is a remarkable position. We have an industry developing technology that its own researchers warn could pose catastrophic risks, while saying that competitive pressures make slowing development extraordinarily difficult. Meanwhile, an NBC News poll found that 81 percent of Americans believe that the government is not doing enough to regulate A.I.
This demands national — and ultimately international — leadership. Congress should move fast to establish enforceable safety standards, independent evaluation of the most capable models and mechanisms for cooperation with China and other A.I. powers.
The technology is advancing at an extraordinary speed. How much more evidence do our political leaders need before concluding that voluntary self-regulation is not enough?
Mervyn Turner Scotch Plains, N.J.
To the Editor:
Our preoccupation with hypothetical A.I. doom feels like a reasonable yet questionable form of displacement. It is far easier to project our anxieties onto unfeeling algorithms than to confront an uncomfortable truth: Our greatest adversary has always been ourselves.
Every year, millions of lives are lost to entirely controllable causes — chronic hunger, treatable illness, armed conflict, car crashes and acts of violence. The tragedy is not that these crises are insoluble; we already possess the resources, medicine and technology to prevent them.
No future superintelligence needs to orchestrate our downfall. We are already authoring it through our own persistent indifference and division. So, while we look to the machines in fear, we must also look in the mirror and recognize that our greatest threat is also our greatest hope.
David C. Zweig Los Angeles
To the Editor:
While I enthusiastically applaud Dario Amodei’s willingness to submit Anthropic’s A.I. programs to third-party review, his company’s new policy doesn’t address the most important question for humanity: Should we allow a private company to own a tool that could cause mass unemployment and economic destruction and threaten lives?
It is illegal for companies to own nuclear weapons and most deadly pathogens, and Congress should make private ownership of superintelligent A.I. illegal as well.
Instead, the United States should contract with companies to develop such a tool. Congress should ensure that a nationally owned superintelligent A.I. system is highly regulated so that it is used solely to protect the American people from attack by our adversaries.
Let’s restore some sanity by allowing private companies to develop only A.I. products that can be deployed safely.
Ilir Zherka Washington
To the Editor:
Not to take the current panic of rogue A.I. too lightly, but Stephen Witt’s essay put me in mind of a 1956 science fiction movie, “Forbidden Planet.” The movie is probably best remembered for its introduction of Robby the Robot, who became a popular children’s toy after the movie’s release. But the heart of the story seems eerily prescient.
On a planet far out in space, an extremely advanced race, called the Krell, develops technology that advances to the point of unlimited power, including the elimination of all disease, death, war and physical labor. But the entire civilization disappeared in one night.
As visitors to the planet millions of years later eventually determine, the Krell’s own technology destroyed their race by releasing the worst impulses of their collective subconscious. Their collective id created monsters that tore them to pieces.
There is much to learn from old movies.
Alice Porter Natick, Mass.
To the Editor:
Let the historical record reflect that it was human intelligence, not so-called superintelligent A.I., that raised the clarion calls to put the brakes on the unthinkable risks of proceeding at dangerous speed with the development of patently foolhardy A.I. models.
William August Cambridge, Mass.
The post The Challenge of Regulating A.I., and Ourselves appeared first on New York Times.
