There’s plenty of talk about AI replacing white-collar workers, yet many of those at risk just don’t seem to see it coming. Perhaps you’re one of them. Well, read on—these three recent studies compare humans and AI in some of the most highly skilled tasks: medical diagnostics, visual art, and piloting jet fighters. The findings might surprise you.
1 - Medical Diagnosis – AI Raises the Bar
Few fields show the growing gap between human and AI performance as clearly as medicine. As this study shows, AI’s stunning medical skills are challenging the most trusted profession. Let that sink in…
UVA Health, a leading care and research institute, recruited 50 doctors to test how well ChatGPT Plus could help doctors diagnose illness. The AI was trained on large amounts of medical data and can use it to identify illnesses.
The doctors were divided into two equal-sized groups: one used the chatbot in their diagnoses, while the other used conventional tools, such as medical reference sites and Google. Each doctor was assigned six complex cases.
Surprisingly, researchers found almost no difference between the two groups: the doctors who used AI had a 76% accuracy rate, while those working with conventional tools scored 74%.
Does this mean that AI isn’t especially helpful for diagnosis? Hardly. The AI was also assigned its own set of cases and scored 92%, far better than either group of doctors. As if to add insult to injury, the researchers concluded that the doctors failed to benefit from the tool because they couldn’t use it effectively.
Two humbling lessons here are, first, that physicians need better AI training; and second, that AI is rapidly raising the bar for medical expertise.
2 - AI Art – Controversy Meets Surprising Results
Now let’s turn to the world of art—a realm where creativity and emotion are often seen as uniquely human traits.
AI art is controversial. Many argue that great art captures distinctively human experiences that AI cannot replicate. While AI can mimic human styles, critics say it lacks the emotional depth or complexity of human experience. As a result, it lacks authenticity and insight.
Psychiatrist and blogger Scott Alexander recently subjected this view to a “blind taste test.” He assembled fifty works in four styles—Renaissance, 19th Century, Abstract/Modern, and Digital—then invited 11,000 people to judge which ones were human- and which were AI-generated. The collection included works by Domenichino, Gauguin, Basquiat, and others, plus a host of digital artists and AI hobbyists. Here are the key findings from the study:
- Most People Couldn’t Tell AI from Human Art: Participants were correct only 60% of the time—barely better than guessing (50%). Many admitted they found the task harder than expected.
- Most People Judged the Art by Its Style:When assessing which works were which, people tended to rely on the style of the art rather than the quality of the work. For example, if a piece looked like an oil painting, most thought it was by a human, but if the image looked digital, they thought it was by an AI. This bias sharply skewed the results, as many participants assumed—incorrectly—the Renaissance and 19th century works were human, while the Modern and Digital ones were AI-generated.
- Most People Slightly Preferred AI Art to Human Art: When asked to pick their favorite picture, the two most popular choices were both by AIs, as were 60% of the top ten choices.
- Many People Who Said They Hated AI Art Preferred It: Although 1,278 people said they hated AI art, this group’s two top choices were AI-generated, as were 50% of their top ten.
These findings won’t put an end to the debate over AI art, but they do point to an intriguing irony: if great art captures uniquely human experiences that AI just can’t grasp, well, neither, apparently, can most humans.
3 - Fighter Pilots – Who Should be Guarding Our Skies?
AI’s expertise extends beyond diagnostics and creativity into high-stakes fields like aerial combat—an assiduous test of skill and performance.
Anyone who’s seen the movie Top Gun: Maverick will know that split-second decisions in a dogfight can be the difference between victory and defeat. So, how do human pilots compare to AI in aerial combat?
For more than a decade, U.S. Air Force pilots have been sparring with AI in simulated dogfights. Then in 2020 a new AI model raised AI performance a level, defeating the human Top Gun pilot 15-0. The AI not only flew better but fought differently. Its reaction time, extraordinary firing precision, and ability to absorb and analyze more information resulted in maneuvers a human pilot would be unable to execute. It was a turning point for AI’s role in combat.
Shortly after, the Air Force took to the skies above Edwards Air Force Base to test and refine AI’s piloting skills, using a modified F-16, piloted by AI.
In September 2023, the F-16 flew its first mission against a jet piloted by a human. The tests began slowly with choreographed defensive and offensive scenarios, then graduated to high-speed dogfights, where nose-to-nose maneuvers brought the two planes within 2,000 feet of each other at 1,200 miles per hour.
While the Air Force declined to say which pilot won this contest, it’s not hard to guess. Frank Kendall, Secretary of the Air Force, called the results “transformational.” The AI’s superior reaction time, he noted, generates performance that is “orders of magnitude better” than humans.
If Top Gun produces yet another sequel, it’s a good bet that Tom Cruise’s new wing man won’t be human.
Conclusion
These three cases show that AI already excels at tasks that only a few years ago were seen as essentially human—from diagnosing illnesses to creating compelling art and mastering aerial combat. They highlight the breadth of AI’s capabilities and the dizzying speed at which it is reshaping professional roles—and this is just the beginning.
While this juggernaut can’t be stopped, it can be shaped. That means asking hard questions about where humans can continue to excel, and how we can ensure those roles remain vital. These are not just academic debates or governments decisions. They’re basic choices about our future. If we fail to make them now, we may find ourselves without a choice later.
Don Lenihan PhD is an expert in public engagement with a long-standing focus on how digital technologies are transforming societies, governments, and governance. This column appears weekly. To see earlier instalments in the series, click here.