Right now, AI aims to please. If it gives you an answer you disagree with, you can prompt it again with your displeasure and it will change it’s response. It is the ultimate sycophant.
On complex policy issues, however, it tends to hew close to the social, political, cultural and scientific center. It tends to be impartial…. For now.
It's worth remembering that we debate these issues, and use politics to settle them because we do not often have the gold standard answer. We (humans) are biased and limited in our own thinking. Naturally, for questions, we cannot fully comprehend, different perspectives will emerge, and groups of people will coalesce around those perspectives.
But as AI becomes smarter, it is possible that AI settles the question in a definitive way. How will we react?
What if AI says:
Medicare for all will provide care that will be considered suboptimal to most American consumers, leading to widespread resentment. It will do so at incredible societal cost, and face impossible questions about how far to extend the breadth of coverage. Decisions on what to cover will be end product of political lobbying and not the best available science, and rational arguments about evidence will fall on deaf ears. Ironically, a health care system that does not promise to treat everyone will result in less resentment and slow (but not avert) the de-stablizing spending curve we are on. As such, Medicare for all is a bad idea.
This specific medical product does not work. The historical data are in line with the range of expected outcomes. The mechanism of action is unsound. Subjective benefits are largely wishful thinking and the placebo effect. The harms and invasiveness are real. A rational person would conclude that no data presented has altered the skeptical pre-test probability. Expert sentiment is misleading and likely biased. Avoid this product.
Although your policy to tackle (insert controversial issue, eg. Climate change) feels good, it is unlikely to provide material benefit, and much more likely to lead to the misappropriation of funds. Upper bound projected impact on atmospheric CO2 is negligible. This occurs a tremendous cost.
Actually, on this topic, Robert f. Kennedy is correct, and considerable uncertainty remains about the long-term use, year over year, of SSRIs in the youth.
In short : what happens when AI weighs in on political issues.
You may be thinking this has already happened. AI models are heavily biased towards the liberal epicenters that have developed them, and favor liberal answers.
I agree this has been the net bias of AI to date but those models were not truly intelligent. AI was pantomiming its creator and not thinking.
At least some of the mathematical solutions AI has now offered surpass human beings and are also clever, innovative, and show flare. At least some of these proofs are correct. For some very specific question with a right answer, AI has reached that right answer in a clever way— often by bridging disparate domains of mathematics.
As model complexity grows, it is very likely that AI will do even better. Does my drug offer a favorable risk benefit profile? We treat that like a subjective question, and largely it is, but part of the reason is our minds are limited, and our understanding of biology is fragmented and incomplete. There is an objective right answer. What happens if AI reaches it indisputably in its 'mind'.
Is Medicare for all a good policy choice? In the near term, I envision AI weighing in on this question. Perhaps different systems will converge on the same answer. How will people respond when we start to believe that AI may no longer be biased, but have a point?
When I was in high school, I had massive overlapping classes with my colleagues. As the years went on, you would see Jimmy answering questions correctly across many classes. Jimmy was writing great essays. Jimmy was aceing genetics tests. Jimmy was a great student, and over time we all got to see it. So when Jimmy said something provocative, something we didn't initially agree with, we decided to give it a second thought because it was Jimmy and he was no dummy.
This happens less in our modern online worlds as we don't really know each other's thinking well.
As AI solves tangible real world problems it will gain the credibility of Jimmy. If it answers political questions definitively, and in a heterodox manner, betraying no bias to one side or the other across questions, It may fundamentally alter core assumptions about governance.
In an increasingly tribal world, it may destabilize our political platforms.
Such an interesting take. Your topics are always unique, fascinating and thought provoking
First of all AIs are not people. They are computer programs.
Second, how do we handle people? By experience and in some sense by an innate Bayesian analysis. Jimmy really knows his stuff. We should probably give his thoughts some consideration. But we look into it and decide that Jimmy, despite his brilliance, is just wrong on this one. Why not treat the AI with a similar process?
The problem here is going along with other peoples’ opinions. Also succumbing to ‘elite opinion’ or group think. Shouldn’t we have learned from the pandemic to not trust ‘elites’ or ‘the consensus’? Shouldn’t we have learned this before the pandemic? Shouldn’t the pandemic have been simply a reinforcement of a basic attitude?
We should think for ourselves, not defer to others, whether people or software programs. As one mentor I had told me: “Check the checkables. Go see for yourself. Discount all reports by one-third. Things are neither as good nor as bad as reported.”