Today I saw a funny example of someone who apparently used AI for research in order to debate a person on a topic. While it sounded very thorough, the issue was that the person seemed to only prompt the AI with the facts he wanted it to know in order to generate the best results, instead of giving it all the facts of the scenario so it could properly assess the situation. By not doing that, as you can imagine, the AI just spat out biased information that favored his points without actually understanding what the counterarguments were. And as you could imagine, once people investigated it, it showed so many holes in his arguments, and it didn’t last.
It makes me wonder, because AI is being used nowadays in business as well. Can you imagine someone with that old-school mentality who doesn’t like the opinions of people who disagree with him, who only wants yes-people, and all he does in a boardroom all day is prompt the AI in ways that will actually agree with his decisions, instead of trying to analyze everything in an unbiased way? That’s a potential danger nowadays too.
So, it makes you wonder whether AI tools in corporations should be restricted to certain people through policies, or at minimum, whether people should be required to disclose that everything was AI-generated and show the prompts and facts that were actually put into the machine to generate the results. It would be an interesting way to keep track records too, because I know nowadays in a lot of places you keep notes about boardroom meetings and so forth.
But it makes you wonder, if you’re integrating things like AI in your business, should you keep detailed notes or, in a sense, quote the prompts to understand exactly how the person got to a certain decision? Because if you accidentally put in a wrong number or only gave it certain information, at least this makes it easier to trace things back afterwards from a business point of view. Something interesting to think about in a world where we’re using more robots and AIs nowadays.