I think it was meant like, which decision had more favorable results. But this might only shift the problem to the entity to decide what is more favorable.

Aside from that, i agree with you, the "right or wrong opinion" mindset is dangerous.

> I think it was meant like, which decision had more favorable results.

How does one even do this? Wait a hundred years? It's still probably impossible to determine the outcome of alternative policies. The right/wrong mindset is not only dangerous, but impossible in principle. But the autocrats won't tell you that.

Wellness and happiness are two markers, if measured correctly/ scientifically that can be of use perhaps.

It depends on how you define "right answers". We could be talking about policy making based on objective truth, which is testable and is grounded in reality.

Given an objective scientific fact, what is the appropriate policy that should be enacted concerning that fact? That depends on your values, priorities and ethical framework, none of which have objectively correct answers.

> which is testable and is grounded in reality.

How? This is absolutely impossible. You can't run trials. Yes, we can analyze the effects of policies (which we should certainly be doing more of) but these results can also take decades. We can only do a singular experiment and so our sample is biased to all hell. There is no objective truth here, only results of a single experiment.

> How? This is absolutely impossible.

It is possible to use GPT-3 to impersonate someone. You give it a profile and it assumes the personality, then speaks from that point of view replicating its biases with uncanny ability. So what they did was to create a simulated poll using GPT-3 by aligning the model to personality profiles distributed identically to the real population.

The result? GPT-3 can predict how an idea would poll, how a population would react to a situation. So you can simulate various policies without actually having to test in reality. You can do MCMC like AlphaGO and see ten moves ahead.

It could probably be used to simulate the stock market and economy as well. Just train the model on the "digital exhaust" of companies to predict their future behaviour in aggregate.

This looks to be one of the most astounding uses of large language models. It would be of immediate interest to politicians, investors, advertisers, music and movie produces or any influencer.

It gives me heavy 'Matrix' and 'Psychohistory' vibes.

https://jack-clark.net/2022/10/11/import-ai-305-gpt3-can-sim...

https://arxiv.org/abs/2209.06899

>You can do MCMC like AlphaGO and see ten moves ahead.

The existence of adversarial attacks shows that most neural networks have pretty bad worst-case performance. Thus sticking GPT-3 into alpha-beta or MCTS could just as easily give you an ungeneralizable optimum, because optimizers are by nature intended to find extreme responses. Call it a Campbell's law for neural nets.

The actual AlphaZero nets are probably more robust because they were themselves trained by MCTS, although they still don't generalize very well out-of-sample: IIRC AlphaZero is not a very strong Fischer Random player.

You're entirely over estimating the capabilities of GPT and other LLMs. They aren't magic.

> policy making based on objective truth

Policy making is not about truth, it is about values and priorities.

I would have a very difficult time coming up with an actionable policy where the objective, implementation and side-effects could all be modeled to a point where the policy in its entirety could be described as based on objective truth.

By "voters who got the 'right answers'", I meant voters whose votes most closely matched the final election results. If some voters always match the results of the group, then those voters are representative of the group psyche and could become official representatives of the group for future elections.