Sitting by a window at Boston's Four Seasons Hotel, overlooking the duck pond in the city's Public Garden, Ray Kurzweil held up a paper showing the steady increase in the power of a computer you can buy for a dollar over the past 85 years.
Neon green lines rose steadily across the page, rising like fireworks in the night sky.
That diagonal line, he said, illustrates why humanity is just 20 years away from the singularity, the long-hypothesized moment when humans merge with artificial intelligence, augmenting themselves with millions of times the computing power our current biological brains can provide.
“Even if you build something thousands or even millions of times more powerful than a brain, you can't predict what it will do,” he said, wearing colorful suspenders and a Mickey Mouse watch he bought at Disney World in the early 1980s.
Kurzweil, a renowned inventor and futurist who has made a career out of making predictions that defy conventional wisdom, made the same argument in his 2005 book, The Singularity Is Near. With the emergence of AI technologies like ChatGPT and more recently efforts to implant computer chips in people's heads, he believes it's time to revise his argument. Last week, he published a sequel, The Singularity Is Near.
Kurzweil is now 76 and moves much slower than he used to, but his predictions have become more acute. He has previously said that he intends to experience the singularity, merge with an AI, and live this way indefinitely. But if the singularity does arrive in 2045 as he claims, there's no guarantee he'll be there to see it.
“A healthy 20-year-old can die tomorrow,” he said.
But his prediction is not as far-fetched as it may have seemed in 2005. The success of the chatbot ChatGPT and similar technologies has led many prominent computer scientists, Silicon Valley executives, and venture capitalists to make bold predictions about the future of AI and how it will change the course of humanity.
Tech giants and other deep-pocketed investors are pouring billions of dollars into AI development, making the technology more powerful every few months.
Many skeptics warn that bold predictions about artificial intelligence could fall apart as the industry struggles with limitations on the raw materials needed to build it: electricity, digital data, mathematics and computing power. In the face of many of the world's problems, optimism about the technology can sometimes feel shortsighted, even arrogant.
“People who say AI is going to solve every problem aren't actually looking at what the causes of those problems are,” says Shazeda Ahmed, a researcher at the University of California, Los Angeles who has investigated claims about the future of AI.
Of course, imagining how human consciousness could merge with a machine is a big leap, and people like Kurzweil have struggled to explain exactly how this might happen.
Born in New York City, Kurzweil began programming computers as a teenager, when they were the size of a room. In 1965, at age 17, he appeared on the CBS TV show “I've Got a Secret” to play a piano piece that he had composed using a computer he had designed.
While attending Martin Van Buren High School in Queens, he exchanged letters with Marvin Minsky, one of the computer scientists who founded the field of artificial intelligence, at a conference in the mid-1950s. He soon enrolled at the Massachusetts Institute of Technology, studying under Dr. Minsky, who became the face of this new academic pursuit that blended computer science, neuroscience, psychology, and an almost religious belief that thinking machines were possible.
When the term artificial intelligence was first introduced to the public at a conference at Dartmouth College in 1956, Dr. Minsky and other computer scientists there thought it wouldn't be long before they could build machines that could match the capabilities of the human brain. Some even claimed that within 10 years, computers would beat the world chess champion and discover their own mathematical theorems.
They were a bit too optimistic: We had to wait until the late 1990s for a computer to beat the world chess champion, and the world is still waiting for a machine to discover its own mathematical theorem.
After founding a series of companies developing everything from voice-recognition technology to music synthesizers, Kurzweil was awarded the National Medal of Technology, the United States' highest honor for achievement in technological innovation, by President Bill Clinton. Kurzweil's profile continued to grow as he wrote a series of books predicting the future.
Around the turn of the century, Kurzweil predicted that AI would match human intelligence by the end of the 2020s, with the singularity occurring 15 years later, a prediction he repeated in 2006 when the world's leading AI researchers gathered in Boston to celebrate the 50th anniversary of the field.
“There was a polite chuckle,” says Subbarao Kambampati, an AI researcher and professor at Arizona State University.
AI began to advance rapidly in the early 2010s, when a group of researchers at the University of Toronto explored a technique called neural networks. These mathematical systems can learn skills by analyzing vast amounts of data. By analyzing thousands of photos of cats, they can learn how to identify cats.
This was an old idea that was rejected decades ago by Dr. Minsky and others, but with the advent of the massive amounts of data the world has uploaded to the internet, and the raw computing power needed to analyze all that data, this idea has started to work in surprising ways.
As a result, ChatGPT was born in 2022, driven by the exponential growth in computing power.
Geoffrey Hinton, a professor at the University of Toronto who was instrumental in developing neural network technology and may have done more than any other researcher to contribute to its success, once dismissed Kurzweil's prediction that machines would surpass human intelligence within 10 years, but now believes it was insightful.
“His predictions don't seem so ridiculous anymore. Things are happening much faster than I expected,” said Dr. Hinton, who most recently worked at Google and has led Kurzweil's research group since 2012.
Hinton is among the AI researchers who believe the technology powering chatbots like ChatGPT could become dangerous and even bring about the end of humanity, but Kurzweil is more optimistic.
He has long predicted that advances in AI and nanotechnology, which have the potential to change the way the human body works and the subtle mechanisms that control disease, will resist the inevitability of death. In the near future, he said, these technologies will extend lifespans at a faster pace than human aging, eventually reaching an “escape velocity” that allows people to extend their lifespan indefinitely.
“By the early 2030s, we will no longer die of old age,” he said.
Kurzweil explained that if we can reach this moment, we can probably reach the singularity.
But Sayyash Kapoor, a Princeton University researcher and co-author of the influential online newsletter AI Snake Oil and a book of the same name, says the trends that underpin Kurzweil's predictions — simple line graphs showing the long-term growth in computer power and other technologies — don't always play out as people expect.
When a New York Times reporter asked Kurzweil in 2013 if he predicted his own immortality, he replied, “The thing is, I can't get on the phone with you in the future and say, 'Yeah, I did it, I lived forever,' because you're not going to live forever.” In other words, what he says will never be proven right.
But he may be proven wrong. Sitting by his window in Boston, Kurzweil acknowledges that death comes in many forms. And he knows his margin of error is narrowing.
He recalled a conversation he had with his aunt, a psychotherapist, when she was 98 years old. He explained the escape velocity theory — the theory that humans will eventually reach a point where they can live indefinitely. His aunt responded, “Can you make it happen faster?” Two weeks later, she died.
While Dr Hinton was impressed by Kurzweil's prediction that machines will be smarter than humans in 10 years, he was less impressed by the inventor and futurist's idea of living forever.
“I think a world run by 200-year-old white men would be a scary place,” Dr. Hinton said.
Audio Producer Patricia Zurbaran.