What will you do that makes your game better than the many other computer games out there?
I presume that random members of the public will be playing this game, not just a handful of experts.
Once the AI realizes its in a game and being tested, you have basically no reason to suspect its behaviour in the game to be correlated with behaviour in reality. Given random people who are making basically no attempt to keep this secret, any human level AI will realize this. Freely interacting with large numbers of gullible internet randos isn’t the best from a safety standpoint either. I mean a truely superintelligent AI will persuade MIRI to let it out of any box, but for a top human level AI, it could more easily trick internet randos.
If you kept the controls much tighter, and only had a few AI experts interacting with it, you could possibly have an AI smart enough to be useful, and dumb enough to not realize its in a box being tested. But this makes getting enough training data hard.
I agree strongly with your point about experts vs laypersons interacting with the AI, Donald. I personally am really excited for the potential of game / simulation-world interactions with varying numbers of AI agents and human experts. But I think this a high risk scenario if you are trying to evaluate a new cutting-edge model to test how dangerous it is. I think this project as stated by Encultured AI has value iff it carefully uses only well-vetted models under careful surveillance, and is made available for evaluating new models only in special high-security settings.
What will you do that makes your game better than the many other computer games out there?
I presume that random members of the public will be playing this game, not just a handful of experts.
Once the AI realizes its in a game and being tested, you have basically no reason to suspect its behaviour in the game to be correlated with behaviour in reality. Given random people who are making basically no attempt to keep this secret, any human level AI will realize this. Freely interacting with large numbers of gullible internet randos isn’t the best from a safety standpoint either. I mean a truely superintelligent AI will persuade MIRI to let it out of any box, but for a top human level AI, it could more easily trick internet randos.
If you kept the controls much tighter, and only had a few AI experts interacting with it, you could possibly have an AI smart enough to be useful, and dumb enough to not realize its in a box being tested. But this makes getting enough training data hard.
I agree strongly with your point about experts vs laypersons interacting with the AI, Donald. I personally am really excited for the potential of game / simulation-world interactions with varying numbers of AI agents and human experts. But I think this a high risk scenario if you are trying to evaluate a new cutting-edge model to test how dangerous it is. I think this project as stated by Encultured AI has value iff it carefully uses only well-vetted models under careful surveillance, and is made available for evaluating new models only in special high-security settings.