At a Research Administrator conference. Session on AI.
Two take aways:
1. If you don't know what an LLM is, and how it works (in high level form), educate yourselves so you can be aware and ask competent questions.
2. AI really can't review grants, but is good at *weaknesses* in applications.
Not a fan of what this publisher represents, but also, legit "no shit" moment for value of the humanities and social sciences: www.forbes.com/sites/pialau...
Observable facts, how quaint!! Who needs empiricism?
This is an important talk, a powerful piece of forensics and political media analysis.
Suddenly, out of nowhere, a declassified World War II-era CIA guide to sabotaging fascism in the workplace has become one of the most popular free ebooks on the internet:
www.404media.co/declassified...
You can take it. Bahaha
Sounds like the history of qualia :D
I am begging the national media to start taking this seriously.
Stop buying the claims that the framing of "temporary" or "pause."
They are altering the review processes of federally funded scientific research from merit-based peer review processes to political screening by ideologues.
WTF does he know about any of those adjectives?
So, is the TL;DR of the DeepSeek paper basically they taught a model how to use critical "thinking" skills rather than just brute force training? Am I missing something here where basically AI engineers realised that AI learning could excel if it used transferable skills like human learners?
Probably. I saw that the other day and almost barfed up my mouth of tea. It was half swallowed. But this...
Finally making the full move over here. Twatter is closed down, and taking a long hard look at Meta apps as well. Reading up on the PDS options.
Hey, finally diving in. Changed my handle
dumping the other place, finally here!