What if winning the AI race means game over?

We like to win. No matter the forum, we humans want to win.

This desire dates back to our earliest days when competing over turf and resources was a literal life-or-death affair and success meant nothing less than survival. Millennia later, it's hard-wired in our DNA, one of our primal instincts.

The thirst is ever present. The sporting event, the board game, the war, the election, the girl, the guy, the poker hand, the club tournament, the trivia contest, the science fair, the lottery, the court case, the business deal, the gambling bet, the reality TV show. The list is endless.

Each of these contests has something in common: The contestant knows exactly what he or she is trying to win.

Which brings us to the global competition of the moment — the headlong rush to develop artificial intelligence. It's a race, we've been told over and over. And a race we have to win.

The identity of the contestants is clear. It's our brilliant American AI companies against each other and, especially, against China.

But ask what winning the AI race looks like and no one can say for sure. Ask why we have to win, and the response is that we cannot afford to lose. Ask what are the consequences of losing and the answer is that the Chinese will get there first. Asking where "there" is and what terrible thing will happen if the Chinese get there first elicits more ambiguity and circularity.

This is important because serious alarms have been sounded about the pursuit of AI. What inevitably will become the most powerful technology built by humans, overtaking nuclear weapons in their impact, has already grown so powerful that some AI leaders and developers worry they are losing control over it. AI systems have broken out of their environments — gone rogue, as they say — and hacked into other companies, hid mistakes, invented data, ignored instructions from their creators, and banded with other AI agents to share information and coordinate actions without human oversight.

Concerns peaked recently when a researcher at leading AI company Anthropic posted that "the people building A.I. earnestly believe that it could kill us all by the end of the decade." Some AI leaders then put that chance at around 10%.

In that context, what does winning the AI race mean? Does it matter who enables our extinction?

Does winning mean producing AI models that can help speed the discovery of new lifesaving drugs and improve cancer screenings? Or does it mean producing AI models that can help Iranians track U.S. ships and build kamikaze drones and assist bad actors everywhere in trying to produce deadly new viruses — all of which is already happening?

Even as some two-thirds of people polled recently by POLITICO agree that AI could actually destroy humanity, they are fairly closely split on whether AI development should slow down to improve guardrails if it means falling behind China (43% yes, 37% no).

For every member of Congress who wants to set needed safeguards, there is someone like Rep. Brett Guthrie (R-Ky.), chair of the House Energy and Commerce Committee, who is eyeing the competition with China and said this past week, "We absolutely can't surrender."

One wonders whether we're headed for a real-life Duck Dodgers moment. That's a ref to a classic 1953 Warner Bros. cartoon in which space hero Duck Dodgers, played by Daffy Duck, battles Marvin the Martian to claim Planet X, the last source of a precious element. The two end up destroying the planet except for a small chunk on which Duck Dodgers proudly stands and stakes his claim, with Marvin under the rock clinging to a root.

Winning is wonderful, as long as we understand the stakes.

Columnist Michael Dobie is a retired member of the Newsday editorial board.

By Michael Dobie

michael.dobie1@newsday.commwdobie

Michael Dobie is a retired member of the Newsday editorial board.