Person
1 episodes · 25 quoted claims
According to Ray Kurzweil, a 1999 Stanford conference polled AI experts on when AI would pass the Turing test, and the average answer was a hundred years.
“It was really the first time that AI experts worldwide were polled on this prediction. Uh, and the average poll was a hundred years.”
In the 1999 Stanford poll, 20 percent of AI experts said a machine would never pass the Turing test and 80 percent said it would happen but not in their lifetimes.
“Uh, 20% believed it would never happen. And that was the view in 1999. 80% believed it would happen, but not within their lifetimes.”
Ray Kurzweil says that a year before the interview, the forecasting platform Metaculus polled AI experts and gave 2042 for passing the Turing test, later revised to 2030.
“So a year ago, uh, something called Meticulous, which you may be aware of, assesses different types of experts on the future.”
Alan Turing described the Turing test in a 1950 paper, with only a few lines on it and a suggestion that fifteen minutes of conversation could suffice.
“the 1950 paper where he had articulated the Turing test, he there's like a few lines that, uh, talk about the Turing test”
Ray Kurzweil says language models worked poorly at around 10 billion parameters, worked fairly well at 100 billion, and were moving toward 1 trillion.
“it requires a certain amount of information in order to, uh, be effective, and that we would not see AI really being effective until it got to that level.”
Ray Kurzweil says Moderna simulated billions of mRNA sequences in two days to design its COVID-19 vaccine, with ten months spent on testing.
“They did that in two days.”
Ray Kurzweil says AlphaFold predicted protein 3D shapes with near-100 percent success, where human experts managed perhaps 15 to 20 percent.
“human beings could do the best humans could maybe do fifteen, 20%, uh, of figuring out what the shape would be, and the this after a few takes it, uh, it ultimately did just about 100%.”
Ray Kurzweil says the first computer was created by a German during World War II, and that the German government rejected it as insignificant.
“the very first computer actually created by a German during World War II. And you might have thought that that might be significant, but actually, the Germans didn't think computers were significant”
Ray Kurzweil says a British computer cracked the German code during World War II, which he credits with helping Britain win the Battle of Britain.
“the third one was the British computer, uh, which the Allies did take very seriously, and it cracked the Ger, the German code”
Ray Kurzweil says exponential growth in computing began decades before Intel and moved from relays to vacuum tubes to transistors to integrated circuits.
“Intel, in fact, this started decades before Intel was even created, and wasn't with transistors formed into a grid.”
Ray Kurzweil says the training compute used for the largest AI models has doubled roughly every six months.
“the fastest computer models that have been created, and that actually expands roughly twice a year, like every six months, it expands by two.”
Ray Kurzweil says Go champion Lee Sedol said he would stop playing Go after a machine went far beyond his ability.
“Lee Sedol, who's like the best Go player in the world, says he's not gonna play Go anymore.”
Ray Kurzweil cites a poll in which 70 percent of respondents thought extreme poverty had worsened and only 1 percent correctly said it had fallen by half.
“70% thought it had gotten worse, and and that's the general impression.”
Ray Kurzweil says life expectancy was 48 years in 1900 and is now over eighty.
“Life expectancy was forty eight in 1,900,”
Ray Kurzweil used cell phones, which started expensive and became affordable, to argue that advanced technology becomes cheap over time.
“I mean, there's like 4 billion cell phones in the world today.”
Ray Kurzweil notes that two nuclear bombs were used in 1945 and none has been used in anger since, which makes him think nuclear war will be avoided.
“We had two nuclear bombs go off in 45, so now we're seventy seven years later.”
Ray Kurzweil describes a product that read 200,000 books and returned the sentence that best answered a question, which he adapted to his father's writings.
“it would read 200,000 books, and then, and and then find the one sentence in 200,000 books that best answered your question.”
Ray Kurzweil says he built a question-answering system from the writings of his father, Frederick, who directed choral and music groups.
“everything that Frederick Israel had written.”
Ray Kurzweil says his great-grandmother founded a school in 1868 that was the first in Europe to give girls higher education, through 14th grade.
“my mother's mother's mother created the school in 1868, and it was the first school in Europe that provided higher education for girls.”
Ray Kurzweil says he won the Westinghouse Science Talent Search as a high school student.
“where I won the Westinghouse Science Talent Search.”
Ray Kurzweil says he developed optical character recognition in the 1970s and waited until the 1980s for speech recognition, which needed more computation.
“I did OCR in the 1970s, uh, because OCR is, uh, doesn't require a lot of computation.”
Ray Kurzweil says his two analyses of the Drake equation in The Singularity Is Near gave either thousands of advanced civilizations per galaxy or just one.
“I have two analyses of the Drake equation, both with very reasonable assumptions, and one gives you thousands of s advanced civilizations in each galaxy, and another one gives you one civilization.”
Ray Kurzweil says he expects longevity escape velocity, where life expectancy rises more than a year per year, by the end of the decade.
“Longevity escape velocity, um, which we're not quite at yet, but I think we're actually pretty close.”
Ray Kurzweil says salmonella from a trip to Mexico caused pancreatitis that left him with an unusual form of diabetes, neither type one nor type two.
“Went to Mexico 40 years ago, developed salmonella that created pancreatitis, which gave me a strange form of diabetes.”
Ray Kurzweil describes the 'great goo' risk as a self-replicating nanobot that could turn everything into copies of itself and destroy biological life.
“it's the great goo challenge.”