A friend of mine who is a computer scientist describes artificial intelligence as a Rorschach test. Invented by the psychologist Hermann Rorschach in the early 20th century, the famous test involves showing subjects a series of ink blots and asking them to describe what they see. The test is a psychoanalytic technique to help determine a psychological profile of the patient.

When my friend calls AI a Rorschach test, he means that what we see in AI says more about us than it does about the AI. I have seen this phenomenon firsthand. I recently participated in a workshop aimed at developing ethical guidelines for research using AI. The workshop participants came from a wide variety of backgrounds: higher ed, K-12 education, data science, mathematics, public policy, and industry. Everyone had some familiarity with AI, but it covered the entire spectrum from people who never use it to people who build and design it.

Surprisingly, the people who had the most familiarity with AI differed wildly in their judgments of its capabilities. Some of them were optimistic—they claimed that AI was already revolutionizing scientific research or business. Some of them were deeply pessimistic, claiming that AI’s hallucination problem was endemic, that it was overhyped, and that whatever practical uses it had would be limited ones.

The different attitudes of the apparent experts reminded me of a much older debate about a much earlier technology. In 1665, Robert Hooke published his famous work Micrographia, which contained illustrations of various specimens that he had viewed under a microscope. In the book, Hooke claims that the microscope will allow us to “discern all the secret workings of Nature.”

His contemporary, the philosopher Margaret Cavendish, was unimpressed. She had her own microscope, and she didn’t think they worked that well. The lenses frequently broke or cracked, and contained other flaws that distorted what you were looking at. Not only that, but what happens if your microscope is in bad lighting? As she put it in her Observations upon Experimental Philosophy, you ended up with “shadows, refractions, reflections” that interfered with your specimen. Given all the problems, Cavendish couldn’t fathom why “this art has intoxicated so many men’s brains and wholly employed their thoughts.”

Hooke and Cavendish both knew how microscopes worked and had used them, so how did they come to such different conclusions?

The difference wasn’t in their knowledge; it was in their emotions. Hooke saw potential in the microscope because he approached it with wonder and hope. He was in awe of the tiny world that microscopes revealed. Cavendish approached the microscope with caution and skepticism. Cavendish believed that science ought to meaningfully improve people’s lives. The mere fact that the microscope made tiny things visible didn’t mean it would, for example, help feed more people who were hungry.

It’s tempting to think Cavendish was just wrong because microscopes ended up becoming an important scientific tool, but that’s because we significantly improved them starting in the 19th century. The version she had did have all the problems she attributed to it. Maybe Hooke gets credit for his foresight, but without the eventual improvements, microscopes could have just remained a novelty item.

Hooke and Cavendish’s disagreement over the microscope came down to how they felt about it. Where Hooke saw its potential over its flaws because of his hope, Cavendish saw its flaws over its potential because of her skepticism. Our emotions likewise explain why we differ so wildly in our judgments about AI, even when we know how the technology works.

Some people see its potential because they are hopeful about what it could do. Could AI, given the right training, help us find a cure for cancer? Some people see the flaws because they are skeptical. Like Cavendish, the flaws temper their hopes about what AI might be able to do.

One important lesson for all of us is that the talk around AI right now is more feeling than facts. But the further lesson is that even when people have the same facts, their feelings make them interpret those facts differently. If we want to make some sense of the AI zeitgeist, we’d be better off paying more attention to our emotions.