Can a single AI model advance any field of science?

Re:Can a single AI model advance any field of science? by raddan on Tuesday April 22, 2025 @05:47PM (#43766636)

I think an important question to ask is whether your scientific task is primarily one of interpolation, or one of extrapolation. LLMs appear to be excellent interpolators. They are bad at extrapolation.

[ Parent | on HN ]

Re:Can a single AI model advance any field of science? by immibis on Tuesday April 22, 2025 @05:51PM (#43766665)

Climate models aren't LLMs.

[ Parent | on HN ]

Re:Can a single AI model advance any field of science? by da_chicken on Tuesday April 22, 2025 @06:58PM (#43767083)

They're also not AI.

It remains to be seen exactly how much a climate model can be improved by AI. They're already based on woefully sparse data points.

[ Parent | on HN ]

Re:Can a single AI model advance any field of science? by philipwhiuk on Tuesday April 22, 2025 @08:48PM (#43767656)

Why are they not AI?

[ Parent | on HN ]

Re:Can a single AI model advance any field of science? by mnky9800n on Wednesday April 23, 2025 @04:35AM (#43769858)

All current neural network based climate models use some kind of transformer architecture. So in that sense they are at least related to LLMs.

[ Parent | on HN ]