Post

I see a bunch of trend following people clutching their monocales in horror at IA's use of claude / LLMs to patch their systems. 1) Grow up. If it's so important to you that only "authentic, heritage human code" be used to fix IA's infrastructure, why don't you donate the extreme $$ needed?

Internet Archive@archive.org

An update from Brewster Kahle on today's outage.

21/10/26, 11:54 pm0 reposts4 likes

John Anderson

@johna.org

2) Unless you are doing net-facing engineering, you can't really understand the sheer magnitude of the vulnerabilities disclosed in the last 24 hours. It would be impossible to address them all by hand in the time frame permitted.

John Anderson

@johna.org

3) An LLM is a *tool*. Are you going to punish a painter for not making his own canvas or paints from raw pigment? LLMs can perform the super-detailed, super-intensive parts of software development that few engineers practice day to day. In this way, it's a force multiplier. Not all AI is slop.