1st regulation: every LLM must show exactly what they used to train it. All the text, all the photos, all the videos, all the art. They must pay the owners
modifying text, audio, photo or video of anyone without their consent is a felony. 1st time offenders get a 10% annual income fine. 2nd time offenders get 1 year minimum.
Doesnt cover surgical ai, avionic ai, financial ai, … Still plenty of ways ai can kill people and disrupt society. Vetting and testing/ validation needs to be carried out on the algorithms and output of these systems.
Make the AI makers and users legally liable for what their AI does. AI should not be allowed to make decisions.
Kind of how there’s speculation that the US used AI to determine where to bomb in Iran and they bombed a school full of little girls. That is a war crime of negligence and dereliction of duty on Pete Hegseth.
1st regulation: every LLM must show exactly what they used to train it. All the text, all the photos, all the videos, all the art. They must pay the owners
But but won’t you think of the shareholders and Zizianssss please please
Even so, they will still hallucinate. Its a flaw in the method.
Worse: it is the very principle on which it works, it just sometimes hallucinates it correctly.
Doesn’t solve the safety issue
Regulation 2
Doesnt cover surgical ai, avionic ai, financial ai, … Still plenty of ways ai can kill people and disrupt society. Vetting and testing/ validation needs to be carried out on the algorithms and output of these systems.
Make the AI makers and users legally liable for what their AI does. AI should not be allowed to make decisions.
Kind of how there’s speculation that the US used AI to determine where to bomb in Iran and they bombed a school full of little girls. That is a war crime of negligence and dereliction of duty on Pete Hegseth.