My interest in Sornette’s work led me to Eth’s work in chaos theory. As I recall he moved to ETH early in their nonlinear dynamics work which, of course, has exploded in the past 20 years.
One of my AI questions may be partially answered in your LLM and Deep Learning discussions. My sense is LLM’s are limited by the depth of their training data notwithstanding their increasingly massive sizes. “Insane” AI LLM answers appear to be similar to the breakdown in “long term” chaotic weather prediction models where “all”initial conditions are not infinitely known. Deep Learning appears to help constrain the lack of infinite information. Yes?
Nice article.
Very interesting article. Thanks. Nonlinear dynamics are fascinating. Are you familiar with Didier Sornette’s work at UCLA and ETH?
Quantifying financial risk in long tails and identifying financial bubbles by adapting his work in earthquake predictions. https://scholar.google.com/citations?user=HGsSmMAAAAAJ&hl=en
My interest in Sornette’s work led me to Eth’s work in chaos theory. As I recall he moved to ETH early in their nonlinear dynamics work which, of course, has exploded in the past 20 years.
One of my AI questions may be partially answered in your LLM and Deep Learning discussions. My sense is LLM’s are limited by the depth of their training data notwithstanding their increasingly massive sizes. “Insane” AI LLM answers appear to be similar to the breakdown in “long term” chaotic weather prediction models where “all”initial conditions are not infinitely known. Deep Learning appears to help constrain the lack of infinite information. Yes?