Can hacker culture survive LLMs?

@evan I think if hackers reject reliance on them, yes. Just plugging things in to an LLM and using the output is like a new variant on script-kiddie. No learning or understanding, just some megacorp-produced algorithm in a hyperscalar data center doing everything for you. Likewise for using LLM-generated code to make software.

But, people just focused on the "number-go-up" aspect of CTFs or a "move fast break things" time-to-ship minimizing dev workflow? Absolutely hooked on LLMs and also not hacker culture

ragectl@hachyderm.io

@evan hacker culture opened the idea of building AI pollution/defense tools when everything on the internet was being slurped into the cooking vats before people started eating the virtual soylent green. So I think it will survive.

evan@cosocial.ca

@kworker so, "yes, but" only if hackers refuse to use them?

malte@radikal.social

@evan Sure has. I guess I wonder where the nitpicking impulse comes from. It is as if this medium amplifies it.

evan@cosocial.ca

@malte it's the Socratic method. You can win every argument by demanding a rigorous definition of every term under question, and exhausting your opponent until they concede whatever you want to make them say.

kworker@kolektiva.social

@evan I'm not sure if I'd say complete avoidance, or refusal to rely on them, or refusal to use them for anything serious.

Clearly other people using LLMs introduces opportunity to hack such systems, and I wouldn't want to foreclose the possibility of local-only LLM models that can be built and used in ethically and ecologically sustainable ways. But I think offloading your thinking to a giant tech corp is definitely not in line with the hacker spirit or culture I understand.

kworker@kolektiva.social

@evan I'm not sure if I'd say complete avoidance, or refusal to rely on them, or refusal to use them for anything serious.

Clearly other people using LLMs introduces opportunity to hack such systems, and I wouldn't want to foreclose the possibility of local-only LLM models that can be built and used in ethically and ecologically sustainable ways. But I think offloading your thinking to a giant tech corp is definitely not in line with the hacker spirit or culture I understand.

kini@maro.xyz

@evan Running your own local LLM, Giving it a bunch of custom tools, figuring out jailbreaks so it will do pen-testing without complaining.

LLMs are extremely fun when you have a hacker mentality.

kworker@kolektiva.social

@evan I'm not sure if I'd say complete avoidance, or refusal to rely on them, or refusal to use them for anything serious.

Clearly other people using LLMs introduces opportunity to hack such systems, and I wouldn't want to foreclose the possibility of local-only LLM models that can be built and used in ethically and ecologically sustainable ways. But I think offloading your thinking to a giant tech corp is definitely not in line with the hacker spirit or culture I understand.

philip@mastodon.mallegolhansen.com

@malte @evan Ah but the problem is each word individually isn’t enough, looking up the definitions of mountain and top may not lead you to the same conclusion as the general consensus definition for “mountain top”, meaning we’ll also need to link to every n possible combinations of words in the post.

/s

golemwire@fosstodon.org

@evan @malte That depends on the definition of "exhausting" >:)

johannab@cosocial.ca

@evan

Yes, but unfortunately at the cost of recreating the sort of factions we've been trying to get out of for many years.

Gonna be a while, and a lot of pain, before we settle on an acceptable and productive moderate middle and start reskilling not just 'hackers' but users/consumers to be more aware of what their software is doing.