aanes_appreciator [he/him, comrade/them]

Welsh, Social Ecology and Jineology enjoyer, and computer programmer.

  • 6 Posts
  • 9 Comments
Joined 2 years ago
cake
Cake day: March 18th, 2025

help-circle




  • It can be. The issue usually lies in having untrusted code contributed by people who may or may not know what they’re doing. Or, worse, maliciously writing code with AI assistance to insert vulnerable code which the human maintainers are less likely to spot because

    1. More AI use = higher quantity of code submissions
    2. More code submitted = more code to maintain and audit
    3. More code to maintain and audit = More chance of human error or reliance on LLMs (which brings us back to 1.)

    I think there’s ways AI can help software engineers accelerate the shitty work and otherwise keep working as before, but there’s big wave of bozos seeing LLMs do one or two cool things and just throwing it all to the sometimes breaks ur code machine. As with all things having some dialectical thinking helps with adapting moreso than blind end-of-history rhetoric from the likes of anthropic et al.