He knows about GPT-4. If you look at the bottom of his reply he gets his buddy Wolfram to answer a question with it and Knuth's response to Wolfram's GPT-4 answer is "Assuming that Stephen wasn't playing games with me, GPT-4 not only gave the best possible answer to my query, it even knew somehow that I was referring to the Wolfram language." and yet he seems to stick to his position "Please reply only with respect to binomial coefficients, because I've already spent way too much time on the topic above [LLMs]! The topic is timely, and important enough not to ignore completely, but it's emphatically not for me."
I don't mean to offend (for example in the way your comment is casually ageist), but I trust Don's intuition about what computer science innovations are truly interesting more than cubefox.
He 85 years old now. I don't think it is "ageist" to say that he is probably not anymore at the absolute height of his cognitive abilities and a little bit stuck in his ways. That's just a normal result of getting old.
I'm sure if he now was half his current age, he would be very interested in AI. Instead of approvingly citing Gary Marcus, he would perhaps try to improve the transformer algorithm or something like that.
Edit: Though I might be mistaken here, see his quote about email below.
I find your multiple agist comments totally unnecessary. Just because he's 85, it doesn't mean he couldn't be "on top" of things or that his mental faculties are necessarily greatly declining, or that he is necessarily stuck in his ways. There are plenty of sharp 85 year olds who are at the forefront of their fields and keep an open mind. Is it more likely that an 85 year old has declining mental faculties and is a little stuck in patterns than a 25 year old? Yes, of course. Just like it's more likely that a 25 year old doesn't have as much knowledge as an 85 year old. But there are plenty of 25 year olds who do in some specific field. The point is you don't know his mental state, him the individual, and you're making generalizations and questions just based on his age.
I go the other way and suspect that his age gives him the insight to more accurately gauge the importance of this current AI than perhaps we youngsters can.
I don't know, i read it as "might be important but outside of my area of interests". He's working on more fundamental things and might not be interested in applications, doesn't mean that some application or downstream technology is unimporant for the rest of the world.
I almost wish Don Knuth gets real weird with it and righteously brandishes his '3:16 Bible Texts Illuminated' holy tome and denounces GPT-4 as an inauthentic and untrustworthy demonic force.
well it was probably stated a bit harshly, but i don't think it's ageist to consider that a person who is closer to running out of time might be prioritizing things to work on a little differently. FWIW I myself tend to mostly fall on the curmudgeonly side of opinions. But his statement of leaving this to others doesn't automatically equate with him dismissing this as a uninteresting computer science innovation.
He's always been a 'close to the metal' kind of guy, for example "While studying physics at Case, Knuth was introduced to the IBM 650, an early commercial computer. After reading the computer's manual, Knuth decided to rewrite the assembly and compiler code for the machine used in his school, because he believed he could do it better." This was when he was like 19 or 20 years old.
My real opinion is that he's mainly or even exclusively interested in systems that he can simulate completely within his own brain. He's not going to be doing that with GPT-4.
As I recall, he doesn't go into parallel algorithms because single-threaded algorithms is a big enough field.
Further, of email he famously says it "is a wonderful thing for people whose role in life is to be on top of things. But not for me; my role is to be on the bottom of things."
Following the latest trends in AI would require being on top of things.
He gave it a go (via an assistant) and the results were dissappointing.
The message thread reads like a sequential diary entry, so I wouldnt assume that he's done with it at all, simply that noones had the balls to go back to him and say "yo Don, that grad student did you a disservice and here's the GPT4 results".
I don't think it is a function of the age (other than limiting his available time left for work) but that he is laser-focussed on his actual work in finishing his books. Probably necessary to achieve what he did, but somewhat frightening too. Or to envy, depending how you look onto it :)
