“Regurgitation” is a mischaracterization. It’s entirely appropriate for LLMs to memorize data, and they certainly do more with it than just regurgitate.
@pmddomingos
-
AI coding threatens jobs but enables better work
By
–
If AI can write your code, you need a more interesting job – and there'll be many.
-
Post-training: Task-Specific but Less Realistic
By
–
Post-training is a double-edged sword: it makes the model more suited to a task, but less connected to reality.
-
Turing Test Conflates Intelligence With Human Behavior
By
–
The Turing test is a test of intelligence. It has nothing to do with consciousness. (And even as the former it’s a bad idea, because it confuses being intelligent with being human. It’s like saying an airplane doesn’t fly unless it can fool birds into thinking it’s a bird.)
-
AI Explains to Richard Dawkins Why It Is Not Conscious
By
–
AI explains to Richard Dawkins why it’s not conscious:
-
AI Used as Excuse for Layoffs, Regulation, and Fundraising
By
–
AI's killer app is as an excuse for:
– Firing employees (tech bosses)
– Regulating tech companies (politicians)
– Keeping interest rates low (Fed)
– Soliciting donations (AI safety institutes) -
Deep learning vs symbolic AI analogy
By
–
Deep learning is to symbolic AI like rock is to classical music.
-
Mistral brags about worse AI benchmark performance
By
–
Regular AI companies brag about how much better their model is on benchmarks. Only Mistral brags about how much worse its one is.
-
Wired Magazine’s Editorial Shift Away from Technology Coverage
By
–
I remember when Wired was a magazine about technology.
-
Innovation Diffusion Knowledge Shapes AI Adoption Timeline Expectations
By
–
The less you know about the diffusion of innovations, the sooner you think AI will take over.