[ExI] Recursive exponential improvement

[John Clark]

*Last week Google introduced Gemini-3 and against all the benchmarks it
easily beat all other AI's. However, its superiority did not last long.
Yesterday Anthropic introduced Claude Opus-4.5 and it easily beat Gemini-3
in the ability to write computer code. One thing I found particularly
interesting, since it started A**nthropic has always had a policy that
before they hired anybody they gave them a notoriously difficult computer
programming test that the applicant could take home and bring back the next
day; they decided to give Claude Opus-4.5 that test and give it a two hour
time limit to complete it. The result was Claude Opus-4.5 got a higher
score on coding ability than ANY human candidate ever had! If that isn't
screaming "recursive exponential improvement" I don't know what could. *

*And to think, some people are still worried about trivialities like the
war on Christmas, illegal immigration, global warming, and the US not
balancing the budget. *


*Claude Opus 4.5 just dropped
<https://www.youtube.com/watch?v=VhuUcVhwqi8&list=WL&index=3>*

*John K Clark    See what's on my new list at  Extropolis
<https://groups.google.com/g/extropolis>*
2df
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20251125/f18a84b1/attachment.htm>