We've known the roundworm's full wiring diagram for decades and can simultaneously record the activity of all of its 302 neurons, but we're still far from understanding how it works.
@pmddomingos
-
Language Models and Misinformation Detection: A Human Limitation
By
–
Why should language models be able to detect misinformation when humans can't?
-
Finiteness in Propositional and First-Order Logic
By
–
It's finite for propositional and function-free first-order logic, for example.
-
Theorem Prover as Dynamical System: Axioms to Deductive Closure
By
–
A theorem prover is a dynamical system. Its initial state is the axioms, and its final one is their deductive closure.
-
All Learners Perform Equally Well Across Datasets
By
–
All learners are equally good on average over all datasets.
-
ChatGPT Cannot Distinguish True from False Statements
By
–
ChatGPT doesn’t know the difference between a true statement and a false one.
-
Ray Kurzweil’s Exponential Predictions: Spectacular Overfitting Failures
By
–
The most spectacularly wrong predictions are made by overfitting exponentials, and Ray Kurzweil is the world champion of it.
-
ChatGPT Self-Contradiction in Single Paragraph
By
–
ChatGPT contradicts itself within one short paragraph:
-
Contrast between fictional and actual artificial intelligence capabilities
By
–
Sci-fi AI: cold, rational, infinitely powerful
Real AI: randomly making stuff up