3 ms·
> can be more powerful _-detectors than _-makers Surely you have witnessed (Generative) Adversarial Networks. It's called "an escalation" (attack vs defense, e
by mdp2021 4y ago
> can be more powerful _-detectors than _-makers
Surely you have witnessed (Generative) Adversarial Networks. It's called "an escalation" (attack vs defense, etc.).
More to the substantive proposal in the post: remember that incidents during surgery are dramatically reduced just by using checklists - you can already reduce incidents through deterministic regulation of action. And you do not require "fuzzy" systems like ANNs for that.
- courseofaction 4y agoI agree, but the thing is, can the escalation potential of a scam which is essentially "send me your money for nothing", with extra steps, confound the understanding of an advanced LLM? The scam has to be both human-comprehensible and plausible in order to be delivered to the victim in the first place, which places a limit on its complexity which the adversarial anti-scam LLM isn't bound to.
- WithinReason 4y agoTo train GANs you need access to the internal states of the generator, which you wouldn't have with these LLMs.
- KRAKRISMOTT 4y agoYou don't need the internal state if you have effectively unlimited API calls querying the generator endpoint. Your loss function can be with respect to the adversarial network only.
- WithinReason 4y agoThen you need an effectively unlimited number of queries to train you network, while the LLM can be trained on a few samples your network produces to counter it.
- KRAKRISMOTT 4y agoWdym? The LLM doesn't have access to your internal state either. Nobody's at an advantage here in a zero sum configuration.
- WithinReason 4y agoJust read what I wrote again.
- KRAKRISMOTT 4y agoYou need to explain your thoughts better.