3 ms·
Given that current LLMs do not consistently output total garbage, and can be used as judges in a fairly efficient way, I highly doubt this could even in theory
by lblume 1y ago
Given that current LLMs do not consistently output total garbage, and can be used as judges in a fairly efficient way, I highly doubt this could even in theory have any impact on the capabilities of future models. Once (a) models are capable enough to distinguish between semi-plausible garbage and possibly relevant text and (b) companies are aware of the problem, I do not think data poisoning will be an issue at all.
- Zecc 1y ago> Once (a) models are capable enough to distinguish between semi-plausible garbage and possibly relevant text https://xkcd.com/810/ https://xkcd.com/810/
- jesprenj 1y agoYes, but you still waste their processing power.
- immibis 1y agoThere's no evidence that the current global DDoS is related to AI.