Can you ask Claude how fast vox day responds between each answer and compared to average human response times for non code related inquiries?
I suspect his speed in reading and comprehension likely will be 5 to 10x faster than normal people which is why the 2.5 SD whereas for most readers maybe you can see 1sd more due to the bottleneck of input plus human processing time.
Watching this amazing collaboration of Vox ands Claude makes me even more disturbed by the Dept of War -- Pete Hegseth and the Prez -- choosing to completely chuck Anthropic out of working with the DoW because, according to several reports, Anthropic refused to 'allow' their AI to make the kill/no-kill decisions.
Ethan Young ("Habitual Linecrosser" online) recently left the Army after 16 years as an 'air defender' -- a missile expert who taught missiles to other senior-ranked enlisted and to officers. He describes, with some angst, that he was involved in a large Pentagon test of AIs as aides in missile defense. AI massively sped up all the work of identifying a putative incoming ... something: friend-or-foe or unknown, likelihood of enemy success, our radar/AI systems targeting speed and precision, and the penultimate decision to shoot or not shoot. Our expected targets are ranked by necessity and ability of successful defense. (We only have so many interceptors, and Ethan points out that 'we' almost always shoot TWO interceptors at each target, based on "probability of kill." But we REALLY don't have a limitless store OF interceptors!
So, "Pawtuckets farm" is important because troops will need food. But the missile factory and a mil-base power plant are likely more important. Ethan agonized that, tactically, a "girl's
school" is NOT a high priority: there are two conflicting ranking methods. Tactical v. moral.
To his dismay and the dismay of most of the other senior enlisted air defenders doing the drills with him -- the AI nearly always chose the tactical ranking. They concluded after all the tests that the AI should not ever be allowed the fire / no-fire decisions. THAT must always be in human hands.
Can you ask Claude how fast vox day responds between each answer and compared to average human response times for non code related inquiries?
I suspect his speed in reading and comprehension likely will be 5 to 10x faster than normal people which is why the 2.5 SD whereas for most readers maybe you can see 1sd more due to the bottleneck of input plus human processing time.
The question is no longer “who are you to challenge the experts?” The question is “is your math correct?”
This quote is gold.
1, 2, 3, 4, 26 million. Math checks out.
Watching this amazing collaboration of Vox ands Claude makes me even more disturbed by the Dept of War -- Pete Hegseth and the Prez -- choosing to completely chuck Anthropic out of working with the DoW because, according to several reports, Anthropic refused to 'allow' their AI to make the kill/no-kill decisions.
Ethan Young ("Habitual Linecrosser" online) recently left the Army after 16 years as an 'air defender' -- a missile expert who taught missiles to other senior-ranked enlisted and to officers. He describes, with some angst, that he was involved in a large Pentagon test of AIs as aides in missile defense. AI massively sped up all the work of identifying a putative incoming ... something: friend-or-foe or unknown, likelihood of enemy success, our radar/AI systems targeting speed and precision, and the penultimate decision to shoot or not shoot. Our expected targets are ranked by necessity and ability of successful defense. (We only have so many interceptors, and Ethan points out that 'we' almost always shoot TWO interceptors at each target, based on "probability of kill." But we REALLY don't have a limitless store OF interceptors!
So, "Pawtuckets farm" is important because troops will need food. But the missile factory and a mil-base power plant are likely more important. Ethan agonized that, tactically, a "girl's
school" is NOT a high priority: there are two conflicting ranking methods. Tactical v. moral.
To his dismay and the dismay of most of the other senior enlisted air defenders doing the drills with him -- the AI nearly always chose the tactical ranking. They concluded after all the tests that the AI should not ever be allowed the fire / no-fire decisions. THAT must always be in human hands.
Can't find the specific bit, but here's Ethan:
~8 min. https://www.youtube.com/watch?v=39SyHekk1Co&t=1862s
10 min https://www.youtube.com/watch?v=BZG04K5_mgo