Artificial Intelligence agents are the new tech hotness. Still, they have a few flaws that not only limit their usefulness but can be weaponized politically and can cause actual harm to people.
One of the biggest issues is what has been characterized as a tendency to “hallucinate,” which is a fancy way of saying that they make s**t up, often in nonrandom ways.
You’ve seen this pop up in stories about legal cases in which lazy lawyers get AIs to write their briefs. It sounds like a nice shortcut because tracking down relevant cases to bolster one’s argument can be time-consuming and tedious, but AIs will often make up cases out of thin air in order to make an argument stick.
AIs make up academic papers, fake news stories, or even invent slanderous stories about people who have become unpopular for some reason or another.
Facebook’s AI has done that to…
