5 Comments

Preach, brother!

Also, look at the front portion of Lana’s computer vision syllabus: http://slazebni.cs.illinois.edu/fall22/

Expand full comment
author

Yes. And I'm going to keep saying that she needs to turn that into a book! Or at least a YouTube series? Featuring you two in the backyard with some cocktails talking about intellectual history.

Expand full comment

Your argument seems transferable to preference learning models (which use permutation statistics). Also, two books that maybe relevant here (both from psychometrics)

+ Foundations of Measurement by Luce, Suppes and others

+ A Theory of Data by Clyde Coombs (from the 60s and from the Michigan school)

Expand full comment

No other way to model language (text and multimodal forms) other than translation into another language (and which is your native one or your creative style) - and moreover transformation is a universal principle of life cycle for any entity - https://www.linkedin.com/pulse/ai-completeness-theorem-solution-michael-molin/

Expand full comment

As it seems everyone forgets that what for humans are texts for machines those text corpora are just sequences of 0s and 1s without any sense. The model of language is created by linguists and takes a lot of human work - https://www.linkedin.com/pulse/general-intelligence-system-michael-molin-2f/

Expand full comment