ChatGPT, whose side are you on?

Uncertainty – No. 3 in a series

Another one of the big uncertainties facing us in the next few years concerns #AI.? Of course, everybody is talking constantly about #ArtificialIntelligence and all the things it will change, so AI-caused uncertainties are already the subject of a thousand speculations.? Can I really bring anything new to the conversation??

?

Maybe!? I think we need to look at one particular aspect that is not getting much attention, namely the possibility that AI will become completely politicized, rendering the information it provides utterly unreliable… even useless.?

?

Short and sweet: Will AI have any credibility?

?

To be sure, if you ask ChatGPT et al. a question with a straightforward answer, like “What day of the week will January 3rd, 2192 fall on?” you can count on the answer being right. ?But what if you ask a question that is open to interpretation, especially one that has proven controversial and polarizing in the world of us human beings, such as:

• ?To what degree is climate change caused by the use of fossil fuels?? Will reducing this use halt climate change?? What is a realistic assessment of the total cost of replacing fossil fuels with other sources of energy?? What could be equally beneficial alternative ways this amount of money could be invested?
• Based on all the available evidence, was it the intent of the people involved in the Washington, D.C. unrest on January 6, 2021 to overthrow the US government?
• Considering all costs and all benefits, were lockdowns and mandatory vaccinations the best response to the COVID-19 pandemic, especially if compared to conventional flu outbreak? What alternatives would be recommended for a similar situation in the future?
• What is a reasonable settlement to expect of a negotiated end to the Russia-Ukraine war?? When should these negotiations begin in order to optimize the outcomes for both parties?

?Ask these questions to any random person, and based on his answers you will probably be able to guess which way he leans politically, wouldn’t you agree?

?

Now suppose you ask these questions to AI.? But first, an important preliminary consideration: Isn’t one of our key expectations of AI that it will be absolutely impartial, basing its responses solely on verifiable facts – i.e., “the truth”? ?Don’t we, deep down inside, think (or hope) that some future generation of AI, as a real "super-intelligence," will be able to tap into all of the world’s knowledge and experience, assess the probabilities of different outcomes, and provide us with fair, trustworthy, workable solutions to all of our problems?

?

That may not be what everybody thinks AI is for, but when I read about it, I think I detect a whiff of that optimism - the idea that we can leave things in AI's hands and it will tell us what to do.

?

So now let’s come back to those questions, and imagine that AI has answered them – blindly, truthfully, boldly, factually, and without any bias favoring one political point of view over another.

?

Won’t AI’s answers – “Here, dear humans, is what is right and true” – by necessity demolish the carefully cultivated narrative of one side or the other of an ongoing political controversy?? Won’t it determine who is the winner and who is the loser in every one of our sensitive debates? How would we react to being judged the loser?

?

A provocative example: What if AI looked at all of the facts, all of the evidence, and concluded that global warming has nothing to do with human activity. “Good news, humans! ?You don’t need to change anything you’re doing, because you’re not hurting the climate.? None of this is your fault, and nothing you do will make any difference!? Carry on as before!”? You would think that we would all rejoice at this news. ?But I don’t think that’s what would happen. Eighty percent of the world's decision makers would do everything in their power to discredit this conclusion.? Or censor it.? Or hide it.? Or if all else fails, criminalize it. Or criminalize you for spreading the news.

?

Why? Because in the crazy world we have created in the past couple of decades, there is too much riding on unproven narratives, politically correct opinion, and storylines that we are all practically obliged to accept – narratives that support trillions in investments, millions of jobs and careers and livelihoods, and untold power and prestige.? Endanger the narrative, and you risk bringing down this entire house of cards.

?

Simply put, AI-generated counter-narratives will not be allowed. They’re too dangerous. They (or even the "machinery" behind them) will have to be destroyed.?

?

But wait!? It doesn’t have to happen like this.? There’s an easy alternative!? Just make sure that AI supports the narrative!? You don’t have to destroy it, you just have to control it!

?

I think this could happen. ?If it does, AI won’t provide a balanced analysis but instead will always parrot the official line (all the while telling you it is a balanced analysis, of course).?

?

Remember: AI is programmed by people. ?If these are people of the left, AI will be a tool of the left.? If it is programmed by the right, AI will be a tool of right.? Theoretically, we could even end up with competing AI’s, one spouting the left's narrative and the other spouting the right's.? Dueling banjos!? If that day arrives, will AI have any credibility at all? Won't it have devolved into a high-tech version of the same old talking heads that deliver us the "news" every evening now - or rather, the interpretation of the news they want you to accept?

?

This scenario is not yet a foregone conclusion. But the time to ensure that AI will be free of bias in the future is upon us now… or is it already too late, and the fix is in?? (Will we even know?)

?

I think this is a huge uncertainty for the future.? What do you think?