… yeah … but brains are great! They’re the nexus of our biological imperatives: 1) survive at all costs; 2) reproduce profusely, and; 3) rear many offsprings — who wouldn’t want a computer that strives to achieve that? q^8

There are apparently four. The MI300P is also on the drawing boards.

I am a bit concerned about the focus on “cognitive simulation”. In my mind, it will be important to not just develop such simulation approach, but also to rigorously evaluate its pros and cons against the more normal methods (ODE/PDE, stats/regression), not just for efficiency and accuracy, but also for explainability. I’ll point, as an example, to Jonathan Resop’s 2006 M.S. Thesis: “A Comparison of Artificial Neural Networks and Statistical Regression with Biological Resources Applications”, where, for the application area, ARIMA models performed as well as ANNs, but with much fewer parameters, and these parameters could be ascribed proper process-oriented meanings. While demonstrative, the development of AI/ML models for their own sake (foregoing comparison with classical strategies), does not contribute very much to knowledge.

I’ll add that if the following statement is meaningful: “the scale and complexity of both experimental and simulated data has moved beyond that which humans can handle in their heads”, then, clearly, we should develop computational aids that are much unlike our “heads”, following on historical trends from clay tablets (for our brain’s inaccurate memory), to the abacus and slide rule (for our brain’s poor math skills), and so forth (i.e. anything but a brain-like device).

These concerns notwithstanding, I do find El Capitan awesome (and Livermore too)!

https://youtu.be/L2KM-E9Ne84