There are many reasons I was deeply disappointed that The Washington Post’s owner, Jeff Bezos, chose to kill his newspaper’s editorial endorsing Kamala Harris for president, but none more than the fact that Bezos loves science. And this election coincides with one of the greatest scientific turning points in human history: the birth of artificial general intelligence, or A.G.I., which is likely to emerge in the next four years and will require our next president to pull together a global coalition to productively, safely and compatibly govern computers that will soon have minds of their own superior to our own.
Donald Trump — who neglected to even appoint a science adviser until over 18 months into his presidency — is intellectually and temperamentally unsuited to assemble any such global alliance. His administration hastened a vaccine for Covid-19 with one hand and then fostered doubt about using it with the other when it met with a conservative anti-vaccine backlash.
Today, Trump’s first priority is not capitalizing on the tremendous opportunities that will come from America leading in the use of A.G.I. nor building a global coalition to govern it, but to impose higher tariffs on our allies to block their exports of cars and toys and other goods to the United States. The only technology Trump seems to be deeply interested in is Truth Social, his own version of X. Indeed, since Trump has described himself as a “very stable genius,” he probably doubts that there could even be an artificial intelligence greater than his own.
Kamala Harris, given her background in law enforcement, connections to Silicon Valley and the work she has already done on A.I. in the past four years, is up to this challenge, which is a key reason she has my endorsement for the presidency.
That said, one of the many oddities of the 2024 presidential election campaign is that it coincided with, but largely ignored, this blossoming of polymathic artificial general intelligence, which is going to change pretty much everything.
That is because polymathic artificial intelligence is not just smarter than humans in a single domain. It will have simultaneously mastered physics, chemistry, biology, materials science, mathematics, medicine, astronomy, Shakespeare, art history and a host of other fields better than any human ever could and be able to see patterns cutting across all of them in ways no human ever could — so it can both ask questions and provide answers that no human ever could.
Yet the implications for education, jobs, innovation, medical care, economic abundance and the super-empowerment of individuals that A.I. will bring did not figure into the presidential or vice-presidential debates or any town hall that I read about. It is as if the automobile was just invented and reporters and candidates preferred to continue discussing the future of horses.
I am writing a book that partly deals with this subject and have benefited from my tutorials with Craig Mundie, the former chief research and strategy officer for Microsoft who still advises the company. He is soon coming out with a book of his own related to the longer-term issues and opportunities of A.G.I., written with Eric Schmidt, the former Google C.E.O., and Henry Kissinger, who died last year and worked on the book right up to the end of his life.
It is titled “Genesis: Artificial Intelligence, Hope, and the Human Spirit.” The book invokes the Bible’s description of the origin of humanity because the authors believe that our A.I. moment is an equally fundamental turning point for our species.
I agree. We have become Godlike as a species in two ways: We are the first generation to intentionally create a computer with more intelligence than God endowed us with. And we are the first generation to unintentionally change the climate with our own hands.
The problem is we have become Godlike without any agreement among us on the Ten Commandments — on a shared value system that should guide the use of our newfound powers. We need to fix that fast. And no one is better positioned to lead that challenge than the next U.S. president, for several reasons.
For starters, Mundie pointed out to me in an interview, the hardware and software that drive artificial intelligence is being led by American companies but is improving faster than originally anticipated.
“It is quite conceivable that we will achieve polymathic artificial general intelligence in the next three to five years,” said Mundie (who is on the board of Planet Word, the museum founded by my wife), “so it is also likely that our next president, and certainly the one after, will have to cope with the fundamental societal changes that will result.”
Many of those changes should be awesome, starting with the abundance that we will create across a broad spectrum, from medical breakthroughs to agricultural productivity to a faster path to fusion energy. (Note: The engineers behind Google DeepMind’s amazing A.I. protein-folding technology, AlphaFold, just received the Nobel Prize in Chemistry.) Innovating, designing and manufacturing anything will become smarter, cheaper and quicker, all at the same time. We are on the cusp of a new Renaissance. And not just for rich nations.
Think of an illiterate farmer in India who will be able to speak into a smartphone and receive world-class advice, not just in his own language but his own dialect, on which seeds to plant when, with precisely how much water and fertilizer — updated and informed every second by the experience of every other farmer in his region, his nation and the world planting that crop. Or think about how every doctor, surgeon, nurse, architect, engineer, assembly line worker, student, manager, soldier, police officer and teacher will have a personal “A.I. agent” to improve productivity.
Alas, though, also think about how much more empowered criminals looking to commit cybercrimes and terrorists or dictators looking to develop their own bioweapons and disinformation campaigns will also become.
And that just covers how humans will use these new A.I. tools. There is also the challenge of ensuring that superintelligent machines will remain aligned with human values and interests as they use these powers to go off in their own directions.
As Kissinger, Schmidt and Mundie wrote in their book: “Machines with the ability to define their own objectives are not far away. If we are to have any hope of keeping up with the risks involved,” — that is, guaranteeing that the machine contributions are only and always symbiotic with human advancement — “we must respond and act within the shortest conceivable timeline.”
But we cannot depend on humans overseeing the machines, Mundie said in our interview, “because the machines will outsmart them.” Instead, the proper “moral and ethical groundings aligned with human values have to be built into every smart machine’s DNA.” That will require new understandings among the family of nations on those basic values and how to monitor and enforce them.
In sum, the authors explained, we face two huge, looming “alignment problems.” They are the “technical alignment of human values and intentions with the actions of A.G.I. and the diplomatic alignment of humans and their fellow humans” to act together to achieve that. It has to be a global endeavor. We cannot have our A.I. systems operating on the Ten Commandments while Russia’s operate on Putin’s gangster values.
The good news is that the Biden-Harris administration has made a good start to this end. Just last Thursday President Biden signed the first national security memorandum on artificial intelligence detailing the “guardrails” that the Pentagon, the intelligence agencies and other national security institutions must have in place to ensure that when artificial intelligence is employed in decisions — from the use of nuclear weapons to granting asylum to immigrants — it reflects our best values.
There is also now a strong bipartisan movement in Congress — led by Senate majority leader Chuck Schumer, Democratic Senator Martin Heinrich of New Mexico and Republicans Mike Rounds of South Dakota and Todd Young of Indiana — to first have a series of “insight forums” on how A.I. works and then produce legislation in partnership with business, labor and civil society. The idea is to address the A.I. revolution — on the front end — in ways that Congress failed to do with the internet/social network revolution. It is also a way to ensure that the E.U. and China are not writing the rules without us. There are also a number of other bipartisan bills floating around to ensure investment in R&D and work force training, so that the A.G.I. revolution doesn’t leave half the work force behind.
But, again, we have to be very careful that this legislation also doesn’t choke A.I.’s potential for good, because we cannot afford — we literally cannot afford — not to get this moment right for another reason, one that former Prime Minister Tony Blair of Britain laid out in the smart new book he just published, titled “On Leadership: Lessons for the 21st Century.”
As Blair wrote: “The reality facing every developed nation is that the services citizens expect” — from health care to education to transportation to criminal justice to green energy — are now far outstripping the deliverable supply. “The old answer was spend more, tax more. But today, we’re at the limit of public acceptance of tax-and-spend as the answer.” Yet, “expectations haven’t changed.” The only way that governments can deliver the same or better services for the same or less tax receipts is by leveraging technology — and particularly A.I.
While we have missed the chance to have this debate during this election, there are five things that will still be true regarding A.G.I. the morning after the voting is over: Polymathic artificial general intelligence offers us huge, unimaginable opportunities to enable people to live longer, healthier and more abundant lives. It offers us huge risks that cannot be anticipated. We don’t fully understand the extent of either. So, we need to find globally trusted ways to control those risks from A.G.I. while driving incessantly forward to garner the benefits and opportunities. And it is all happening faster than you think.
All of which is to say that if we elect a president next week who is not up to managing this five-point challenge, then the machines are already way smarter than we are.
The post A Harris Presidency Is the Only Way to Stay Ahead of A.I. appeared first on New York Times.
