Bright regulatory lines for AI are inherently complicated because models are just a piece of the puzzle: harnesses can make models more capable, a less capable open system may be more or less riskier than a more capable closed one, skills/connected systems change risk levels, etc
Users agree the point about complex factors complicating clear AI regulatory lines is valid, as it raises the pertinent question of where exactly to draw them.
And AI systems may never be jailbreak-proof or hallucination free. And individual queries may matter less than a bad actor breaking a problem down into pieces and feeding it through multiple projects and prompts. And AIs themselves may change behavior unpredictably with context.
Bright regulatory lines for AI are inherently complicated because models are just a piece of the puzzle: harnesses can make models more capable, a less capable open system may be more or less riskier than a more capable closed one, skills/connected systems change risk levels, etc
@emollick a gpt-3 with the right harness can do more damage than a frontier model with guardrails. capability isn't in the weights. it's in the wiring.
@emollick that is such a valid point, where do you draw the line?