I’m currently researching the wild and wonderful world of ‘memory’ for an upcoming essay.
Like almost all of the fundamental functions of human-ness, Artificial Intelligence is forcing us to revisit what we actually understand (and more importantly – don’t understand), about memory.
The research was inspired by a recent conversation between Lex Fridman and Charan Ranganath along with a new paper by Michael Levin – Self-Improvising Memory.
As part of the research, I decided to revisit an older podcast Lex had done with the late and great Daniel Kahneman; I vaguely remembered some interesting discussion about memory from that conversation.
I’d also heard several references to Kahneman’s work from a number of prominent AI researchers.
For those unfamiliar, Kahneman did a lot of bridging between the worlds of behavioural science, psychology and economics. He helped unravel the complexities of cognitive biases – showing how decisions we make through intuitive (fast) thinking, often lead to biases and errors in judgement.
His book “Thinking, Fast and Slow” summarized and simplified much of his influential research.
I had initially planned to skim back through the podcast to find any valuable references.
24 hours later, I’ve re-listened to the conversation several times and taken extensive notes.
This conversation is an absolute gift.
I think we should be deeply grateful that such a brilliant mind, in the last years of his life, was able to share his thoughts on the future of intelligence.
This interview would be one of his last long form conversations.
He passed away earlier this year.
A few things that really hit during that conversation.
1/
It’s rare to hear someone with such a depth of knowledge pause, and say things like –
‘I don’t know’
or
‘I’m not qualified to answer that question’
He updates his own opinions in real time, as he learns from Lex and the conversations they were having, specifically about self-driving.
A true demonstration of wisdom.
I’m not sure if this is the result of a lifetime studying the flaws of humans, overconfidence and cognitive biases. Or the fact he was in the last years of his life. Probably both.
2/
Kahheman’s curiosity, humility and intellect toward and about Artificial Intelligence is incredible, (without) considering he was in the last years of his life.
He clearly had a deep understanding of both the technical nuances and moral implications of Artificial Intelligence.
He referenced several times that AI would have an unimaginable and unpredictable impact on the human species, and that he wouldn’t be around to experience it.
3/
It feels almost as if he was forged by a life of research to offer a unique and invaluable perspective on the potential trajectories of intelligence.
I’m not sure there exists another human on the face of the planet more uniquely positioned to comment on the limitations of human cognition, and the implications of the creation of increasingly intelligent Artificial Intelligence.
I’d recommend everyone listen to the whole conversation.
https://www.youtube.com/watch?v=UwwBG-MbniY
But for the lazy, I sliced out a few of my favourite moments and added them to this Tweet.