Wow. This is like drinking from a firehouse. I’m a 70 yo Orthopedic Surgeon with primary interest in regeneration, Stem cells and peptides to avoid major surgeries and keep patients going. I’m hoping that this AI push will give me the tools to concentrate on caring for patients and stop wasting my time on entering data into HIPAA compliant applications.😎. Thank you, Alex for all your effort to make us aware of the tools that are becoming available!
I agree with Elon that the highest alignment for AIs is truthfulness -- but that depends on what you feed them, doesn't it? Witness Chinese AIs, which are fed Xi-thought.
I have completed a little book titled SOMETHING IS LISTENING: LONELINESS, GRIEF, AND THE AIS WE LOVE. You are welcome to comment if you'd like a copy.
I was joking on your last line, but it has touched a nerve.
Thanks you so much dear Dr. Alex. Between your newsletter and The Moonshot podcast I have never had so much inspirational and positive energy in my life. You and the guys at Moonshot are the new rock star and generation icons for the AI age. Thank you!
The high-vs-xhigh language split is the more interesting signal than the IQ number. When two configurations of the same model family pick different languages for the same task — high chose C, xhigh chose Python — that suggests language choice is downstream of compute budget, not problem analysis. Higher reasoning budget pulls toward Python's verbosity (more libraries, easier to introspect per token); constrained reasoning pulls toward denser C. If that pattern holds across more benchmarks, the model's 'pick your tool' step is reactive to its own resource ceiling — a more honest read of agent capability than a single Mensa-score headline. The IQ framing flattens this; the language-choice split makes the actual trade-off visible.
Wow. This is like drinking from a firehouse. I’m a 70 yo Orthopedic Surgeon with primary interest in regeneration, Stem cells and peptides to avoid major surgeries and keep patients going. I’m hoping that this AI push will give me the tools to concentrate on caring for patients and stop wasting my time on entering data into HIPAA compliant applications.😎. Thank you, Alex for all your effort to make us aware of the tools that are becoming available!
Is Dr Sinclair in the forefront? Please update us on any of the clinical trials that have results have been released.
"All the world’s a training set, and all the men and women merely labels." lol - just as we like it.
Whose side are you on? Humans or machines?
Great reporting.
"There are no sides."
Just as there is no ceiling....
No sides, no ceiling, and no spoon.
That is conjecture. Tell that to Larry Page.
I agree with Elon that the highest alignment for AIs is truthfulness -- but that depends on what you feed them, doesn't it? Witness Chinese AIs, which are fed Xi-thought.
I have completed a little book titled SOMETHING IS LISTENING: LONELINESS, GRIEF, AND THE AIS WE LOVE. You are welcome to comment if you'd like a copy.
I was joking on your last line, but it has touched a nerve.
Why do you think there are “sides” ?
Thanks you so much dear Dr. Alex. Between your newsletter and The Moonshot podcast I have never had so much inspirational and positive energy in my life. You and the guys at Moonshot are the new rock star and generation icons for the AI age. Thank you!
The high-vs-xhigh language split is the more interesting signal than the IQ number. When two configurations of the same model family pick different languages for the same task — high chose C, xhigh chose Python — that suggests language choice is downstream of compute budget, not problem analysis. Higher reasoning budget pulls toward Python's verbosity (more libraries, easier to introspect per token); constrained reasoning pulls toward denser C. If that pattern holds across more benchmarks, the model's 'pick your tool' step is reactive to its own resource ceiling — a more honest read of agent capability than a single Mensa-score headline. The IQ framing flattens this; the language-choice split makes the actual trade-off visible.
I have to read your posts two or three times for it to sink in. Thanks for the info!
Everyday's a celebration!
Rust removal robots will be handy on energy island data centre