Essays/linkedin/24-03-23 ai for everything

First published:

Last Edited:

Number of edits:

­čĄľ If the science is not there, AI will not be a magic wand that solves all problems. My bets in the near future are on Raman Spectroscopy, but I am skeptical of most proposals: Noise-in, noise-out.

What OpenAI seems to have installed in all our minds is that if you feed AI with enough data, you can get whatever you need out of it.

Feed it with text, and you get a writer, feed it with images and you get a painter. Feed with with papers and you get a scientist.

However, books are written with a limited set of words, words with a limited set of characters. Feed an AI just a dictionary and you won't get much out of it. Feed it the ASCII table and you won't get anything out.

I have come across many companies that are pitching the idea of throwing AI to whatever data they generate to get information that eluded many before them. Spectroscopy is a fantastic candidate: 1-D data, mostly cheap instrumentation. But, noise-in, noise-out.

By all means, I don't want to underplay the role of AI. I just think it would be wise for founders, and people doing tech due diligence, to never lose sight of the fundamental aspect: what is the data source. Are you ever be able to give a result with a confidence higher than flipping a coin?

AI algorithms are moving at a fantastic pace. Experimentalists barely have time to catch up with them. Now that both branches are meeting, it'll become a very interesting time to see what happens.


These are the other notes that link to this one.

Nothing links here, how did you reach this page then?


Share your thoughts on this note
Aquiles Carattino
Aquiles Carattino
This note you are reading is part of my digital garden. Follow the links to learn more, and remember that these notes evolve over time. After all, this website is not a blog.
© 2021 Aquiles Carattino
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License
Privacy Policy