Will Artificial Intelligence Kill Us All?
04/08/2023
A+
|
a-
Print Friendly and PDF

Everybody is worked up over artificial intelligence. Is it a threat to humanity? Will the robots decide to kill us all like in Terminator? Is AI the explanation for the Fermi Paradox?

Elon Musk wants a pause on AI development. Eliezer Yudkowsky says in Time:

Pausing AI Developments Isn’t Enough. We Need to Shut it All Down

… Many researchers steeped in these issues, including myself, expect that the most likely result of building a superhumanly smart AI, under anything remotely like the current circumstances, is that literally everyone on Earth will die. …

Here’s what would actually need to be done:

The moratorium on new large training runs needs to be indefinite and worldwide. There can be no exceptions, including for governments or militaries. If the policy starts with the U.S., then China needs to see that the U.S. is not seeking an advantage but rather trying to prevent a horrifically dangerous technology which can have no true owner and which will kill everyone in the U.S. and in China and on Earth. If I had infinite freedom to write laws, I might carve out a single exception for AIs being trained solely to solve problems in biology and biotechnology, not trained on text from the internet, and not to the level where they start talking or planning; but if that was remotely complicating the issue I would immediately jettison that proposal and say to just shut it all down.

Shut down all the large GPU clusters (the large computer farms where the most powerful AIs are refined). Shut down all the large training runs. Put a ceiling on how much computing power anyone is allowed to use in training an AI system, and move it downward over the coming years to compensate for more efficient training algorithms. No exceptions for governments and militaries. Make immediate multinational agreements to prevent the prohibited activities from moving elsewhere. Track all GPUs sold. If intelligence says that a country outside the agreement is building a GPU cluster, be less scared of a shooting conflict between nations than of the moratorium being violated; be willing to destroy a rogue datacenter by airstrike.

Frame nothing as a conflict between national interests, have it clear that anyone talking of arms races is a fool. That we all live or die as one, in this, is not a policy but a fact of nature. Make it explicit in international diplomacy that preventing AI extinction scenarios is considered a priority above preventing a full nuclear exchange, and that allied nuclear countries are willing to run some risk of nuclear exchange if that’s what it takes to reduce the risk of large AI training runs.

Personally, I don’t have an opinion. What I’m reminded of is the 1940s confusion over what to do about atomic bombs.

Robert Heinlein’s 1941 sci-fi short story “Solution Unsatisfactory” (finished on 12/24/1940) about the political implications of the invention of radioactive weapons of mass destruction in 1945 (although radioactive dust rather than bombs) might be worth reviewing.

Heinlein was getting inside information from a Berkeley physicist friend who was telling him that Lise Meitner’s discovery of fission with her nephew Otto Hahn, announced in Nature in February 1939, meant that radioactive weapons were a very real possibility.

I consider “Solution Unsatisfactory” to be Heinlein’s most jaw-dropping prediction of the future.

From Wikipedia’s synopsis:

[Colonel] Manning [the General Leslie Groves character in charge of America’s atomic project] hears of fish dying in Chesapeake Bay where the by-products of Dr. Estelle Karst’s [Lise Meitner] research into artificial radioactive materials are being dumped. She was a laboratory assistant of Otto Hahn, the first man to characterize induced fission in uranium, and fled Germany “to escape a pogrom”. Karst is working on radioactive materials for medical uses, but Manning sees its potential as a radiological weapon. …

Manning seriously considers ordering that all people aware of the secret, including himself, be put to death and all records destroyed. He rejects that course because someone else, perhaps German or Russian, is certain to rediscover it. Instead, Manning in 1945 convinces the President [Fiorello LaGuardia] to use the dust against Germany. Since America is officially not in the war, the Americans give the dust to Britain but at the price of the British accepting a complete US ascendancy in the postwar world.

… RAF bombers scatter the dust over Berlin and leave no survivors. The Nazi regime collapses and the new government surrenders.

Manning warns the Cabinet of the great dangers of the new situation, introducing the concepts of the nuclear arms race, mutual assured destruction, and second strike capability. He convinces the President and Cabinet that the only solution is to use the American nuclear monopoly while it still exists. Any other world power, such as the Eurasian Union, might create such dust and bomb the United States within weeks. Still a congressman, Manning convinces the President that there is no time to get Congressional approval and that the Constitution must be bypassed.

The United States issues a “Peace Proclamation” which essentially demands the immediate and unconditional surrender of the rest of the world. All other states are required to disarm and to hand over all long-range civilian and military aircraft, since any airplane can spread the dust. The prohibition on commercial airlines applies to America also; the Army will manage any required civilian air travel. Most of the world complies.

The Eurasians [e.g., the Soviets] did invent the dust for themselves as Manning had warned, and launch a surprise attack. The American victory in the “Four-Day War” owes much to Manning, who had arranged for Congress and President to be outside Washington ahead of the attack, and false rumors of plague to empty New York; nonetheless, 800,000 are killed in Manhattan alone. Eurasian documents completely vindicate Manning’s unconstitutional policies; had the President waited for congressional approval, America would have lost the war.

Manning becomes lifetime head of the new Peace Patrol, with a worldwide monopoly over the radioactive dust and the aircraft which can deliver it. He opens schools for the indoctrination of cadet patrolmen from any race, color, or nationality. They will patrol the sky and “guard the peace” of any country but their own, and would be forbidden to return to their original country for the entire duration of their service; “a deliberately expatriated band of Janizaries, with an obligation only to the Commission and the race, and welded together with a carefully nurtured esprit de corps.”

Manning does not have time to complete his original plans for the Patrol. In 1951, the President dies in a plane crash; his isolationist successor [Robert Taft?] demands Manning’s resignation and intends to dismantle the Patrol. As Manning argues with the President, planes loaded with radioactive dust and piloted by non-Americans appear overhead. Manning is willing to kill himself and treat the capital of the United States as he would treat any other place which he perceives a “threat to world peace”. He wins the standoff and becomes the undisputed military dictator of the world. DeFries [his chief of staff who is the narrator] (himself dying from radiation poisoning) doubts that Manning, now the most-hated man on Earth, can succeed in making the Patrol self-perpetuating and trustworthy. There is no way of knowing how long Manning will live, given his weak heart. The narrator concludes:

“For myself, I can’t be happy in a world where any man, or group of men, has the power of death over you and me, our neighbors, every human, every animal, every living thing. I don’t like anyone to have that kind of power. And neither does Manning.”

And yet, Heinlein didn’t quite get right what would happen. Granted, he probably thought things through better than anybody else did in 1940. But predicting the future is really hard.

But, still, nuclear weapons were simpler to strategize about since they were destructive, while much that AI can do is constructive.

[Comment at Unz.com]

 

Print Friendly and PDF