Anthropic's Disproof of the ETH
26 points
1/21/1970
2 days ago
by bananaflag
Comments
throw310822
2 days ago
eterevsky
Looks suspicious, but the proof is significantly different.
a day ago
throw310822
It's interesting because the simultaneity proves how easy it was. Irrespective of whether they were independently working on the same problem, or someone in one company alerted the other, the fact that they arrived at it together can only mean that they just had to point their model to the problem to have it solved (they say it costed 14M tokens which is a nothing at API prices).
a day ago
mmh0000
14M tokens is far less than I expected.
I've burned 14M tokens just having Claude build a Grafana dashboard.
a day ago
adriencloud
[dead]
9 hours ago
tanseydavid
Newton meet Leibniz. Alexander Graham Bell meet Elisha Grey. Fable meet Astra.
a day ago
fwip
Wonder if they were editing the same wiki.
a day ago
kauai1
[flagged]
a day ago
So Anthropic claims that Claude disproved ETH on October 16, the exact same day OpenAI did it.