This piece
https://www.1843magazine.com/features/deepmind-and-google-the-battle-to-control-artificial-intelligenceby Hal Hodson, tech correspondent for
The Economist, has quotes I gave him in a phone interview. For the sake of historical accuracy, I thought I'd clarify details. (I didn't have approval of the piece. If you don't count the Korean cable TV interview during the Yi Sedol-AlphaGo match, this is the first time I've gone on the record with the media about Demis and the 1990s stuff. Worth getting right somewhere.)
"Charles Matthews, the Cambridge Go master who ran the tournament, remembers the expert player’s shock at being thrashed by a 19-year-old novice." That was Jack Sun, Chinese 3 dan around Cambridge in the mid-1990s. It was the Sonoyama Tournament, played in Trinity College. That means that it was on 13x13 boards, played with handicaps as stated, here of nine stones given the grade difference, ten minutes each on the clock. I remember Jack going outside to smoke rather often during the evening, and he was certainly outraged that the handicaps made for a fair game, an attitude other stronger players have been known to share.
"Matthews believes that Hassabis founded the company simply to gain managerial experience." I saw a bit of Elixir Studios, when my son did work experience there.
"When Hassabis met Masahiko Fujuwarea, a Japanese board-game master, he spoke of a plan that would combine his interests in strategy games and artificial intelligence: one day he would build a computer program to beat the greatest human Go player." This takes a bit more unscrambling.
Fujiwara Masahiko is a Japanese mathematician and writer whom I met in Cambridge in 1988-9. He played shogi as a student, as well as being a go
shodan, and when Demis travelled to Japan in 1997, as stated, I gave him contact information because of Demis's interest in shogi. They met up, and Demis played shogi against Masahiko's wife's nephew. I was in Japan a couple of years ago, and met up with the Fujiwaras. Masahiko told me on that occasion that Demis had made such comments 20 years ago. Demis, it has to be said, didn't recall that when I tried it on him.
"Traditionally, a student of Go pays back their teacher by beating them in a single contest. Hassabis thanked Matthews by beating the entire game." This is a paraphrase of my joke about the whole business: "In East Asia one repays the teacher by defeating him or her: in Cambridge, it seems, it is enough to write software to do it for you."
Hodson was put in touch with me by Ross Anderson, professor at the Cambridge Computer Lab, whom I met many years ago at the go club. At one point of the interview he told me I had given him seven Demis anecdotes he'd not heard before. I deduce that most of the hacks did their DeepMind research by reading press cuttings (a retro way of putting it). I'm quite glad to have had some of those stories read into the historical record.