roman

@hi@romanzolotarev.com


home: ✔ https://romanzolotarev.com

87 following, 558 followers

2 ★ 0 ↺

[?]roman » 🌐
@hi@romanzolotarev.com

just read ignore the slop by daniel stenberg

what are your thoughts on using ai to discover vulnerabilities or assist with research in general?

...

[?]mirabilos🐈‍⬛ » 🔓
@mirabilos@toot.mirbsd.org

@hi such use is still using the fashtech thingy while burning down the one planet we have and accepting thievery, slavery and abuse

...
0 ★ 0 ↺

[?]roman » 🔓
@hi@romanzolotarev.com

good points. i wonder what alternatives we can offer people who want to do research and discover bugs without relying on big tech....

train their own models ethically and run them locally or simply hire experts to do the actual work?

    ...

    [?]mirabilos🐈‍⬛ » 🔓
    @mirabilos@toot.mirbsd.org

    @hi

    train their own models ethically and run them locally

    doesn’t work, they all use pretrained fashtech models as a base, lest they be too limited for what the sloperators want, and training anything up from scratch is ecologically untenable, and it’s still mind rot

    0 ★ 0 ↺

    [?]roman » 🔓
    @hi@romanzolotarev.com

    and what to do with vulnerabilities discovered with llms? do we fix the code based on that? we can't ignore it,.. right?

      ...

      [?]mirabilos🐈‍⬛ » 🔓
      @mirabilos@toot.mirbsd.org

      @hi we can refuse to interact with the slop extruder directly, and we can “not encourage people to use the fashtech”; beyond that… iff you even get the people to respond themselves… it’s up to the individual

      (practice shows that the noise to signal ratio makes it not worth it anyway)

      [?]4 mice in a php file [he/him (they/them works too)] » 🔓
      @algernon@come-from.mad-scientist.club

      @hi @mirabilos based on recent vulns hyped up by ai companies that ended up being just marginally more than a nothingburger... I think we can safely ignore most of them just fine.

        [?]💙🩷💜 Brett [he/him/any] » 🌐
        @brettm@swarm.coiloptic.org

        @hi@romanzolotarev.com i am anti-ai but if someone has discovered a bug with ai we can't ignore it, as 10 million other people can now easily find out about the same bug

          History