The arms race in artificial intelligence took a surprising turn in January when the Chinese company DeepSeek released a chatbot that appeared nearly as proficient as bots produced by American tech firms, but cost very much less. The release prompted a selloff of the stock of Nvidia, which has a stranglehold on the expensive computer chips previously thought essential to AI. History-minded pundits proclaimed a new “Sputnik moment,” referring to the 1957 Soviet launch of a satellite of that name before America managed to get a satellite of its own into orbit.
The DeepSeek news also prompted predictions that the next breakthrough in AI — the achievement of artificial general intelligence, or AGI — was closer than recently thought. The chatbots to date are good at producing text and images. Specialized algorithms can play chess and solve math problems. But more complicated tasks of the kind that come naturally to humans have largely stumped them. DeepSeek suggested that AI evolution was proceeding more rapidly than anticipated, and that AGI was only a year or two or three away.
One characterization of AGI is that it will be able to do anything a human with a computer can do. Another version is that it will do anything a human with a brain can do. It won’t mow your lawn or paint your house — but it might design devices that will.
The predictions may be premature. A dozen years separated Sputnik from the Apollo 11 lunar landing. But few knowledgeable people are willing to say AGI won’t ever come, and most expect it within a decade at most.
What effects will it have on our lives?
The same reflex that provided the Sputnik analogy to DeepSeek suggests a scan of the past for comparably revolutionary technologies. Agriculture changed human lives more than AGI is likely to. It threw more than ninety percent of the human workforce out of jobs as farming replaced hunting and gathering. It triggered a massive increase in the human population. And it made societies more susceptible to famine and disease than they had ever been.
But the agricultural revolution took centuries and millennia. Individuals and societies had time to adapt. AGI is coming at us in a rush. And once AGI turns its talents upon itself — bots designing bots — the rush will accelerate further.
The printing press, too, revolutionized human life more than AGI is likely to. Before Gutenberg, communication was mostly one-to-one or one-to-several and mostly in real time. Writing had been around for millennia, but it was painstaking to produce and hence touched the lives of only the very few. Printing reduced the cost of production of writing to a small fraction of what it had been, thereby expanding the reach of a writer many times in the present and the future.
Compared to the agricultural revolution, the printing revolution occurred in the blink of an eye. It was enormously disruptive to societies. Printing triggered the Protestant Reformation and its generations of wars. Printing made possible the Enlightenment and the political revolutions that followed.
The invention of steam engines inaugurated the industrial revolution, which produced the modern world in which we still live, with its unprecedentedly high standard of living and its continuing derangement of systems of biology and climate that took millions of years to evolve.
The electrical revolution that started in the late nineteenth century touched individual lives — in homes and offices, as opposed to factories and mines — more specifically than the industrial revolution had. It eventually spawned the computer revolution of the late twentieth century, which in turn brought us to where we are with AI now and AGI prospectively.
Each of the new technologies caused upheaval in economies and the societies of which those economies were a part. Many individuals caught in the upheaval never recovered. But societies adjusted and civilization survived – and ultimately thrived.
AGI will certainly cost people their jobs. But just as the children of hunters learned to be farmers, so the children of those displaced by AGI will learn to be something else.
Really? What? Will anything be left that humans can do better than AGI?
Almost certainly yes, although it’s hard to know in advance what that will be. When electricity threw lamplighters out of work, no one knew what the lamplighters’ kids would do. The jobs the kids wound up in hadn’t been invented. Yet never has a large pool of human labor gone unutilized for long.
But what if AGI hogs all the good jobs? What if all the lawyers and doctors and engineers are bots? And humans are relegated to changing linen in hotels and greeting customers at Walmart?
That’s not likely. At any given time there will be tasks humans (perhaps aided by AGI) can do better — that is, more economically — than AGI alone. We humans are a social species. We like associating with our kind. We won’t be happy interacting with robots exclusively.
A doomsday scenario described by some observers of the AI and AGI phenomena has the bots becoming our overlords. Already Elon Musk’s efficiency cadres are said to be using AI to reduce government waste. The human cost to employees fired might be given little or no weight in the AI algorithms. If Musk is still around when AGI arrives, he might do away with the human cadres and turn over decisions to the superbots.
AI is already employed on the battlefield. No army has yet admitted to delegating life-or-death decisions to bots. But with AGI the temptation might become irresistible — if only from an assumption that the enemy will use it against us if we don’t use it against them first.
Would a bot given the task of reducing municipal power consumption shut off electricity to people behind on their payments, and freeze some of them during the winter? Might a bot managing a political campaign plant false stories in the media about opponents? Could a bot designed to discover bank fraud become a money launderer itself?
These are all possibilities. Every technology produces collateral damage. Reliance on agriculture killed many millions from famine. The wars of the Reformation killed millions more. Millions currently die in traffic accidents each year.
Every past revolution had winners and losers. The AGI revolution will too.
Professor I have a question. I was reading an article in the Wall Street Journal awhile back entitled: “The Guy Behind DeepSeek Blurbed My Book in China.” By Gregory Zuckerman. The book he was referring to was “the man who solved the market.” What I find really fascinating about these is the man behind DeepSeek wrote the preface for the Chinese translation of “the man who solved the market” without getting Zuckerman’s approval. How is that possible?