AI today can write our class essays, plan our meals and draft our emails. But not everyone agrees on whether this type of help is good or bad. Recent headlines have warned that artificial intelligence will not just make us lazy and stupid; they’ve suggested AI could lead to the downfall of civilization. But the truth may lie somewhere in the middle of those best- and worst-case scenarios, says Trent Cash. He’s a behavioral scientist at University of Waterloo in Canada.

The type of AI tool and how it’s used make all the difference in whether someone hangs onto or picks up skills, new data show.

AI will most likely weaken our skills when it simply does the work for us. “If we offload a specific skill to AI,” Cash says, “we’re probably not going to retain that skill particularly well.” Staying engaged while using AI — by learning from its feedback or examples — may, however, enhance our skills.

Fears of AI whittling away human intelligence like some sort of digital pocketknife are likely overblown, Cash says. He dismisses the idea that AI will cause people to completely forget how to think.

“Humans are resilient,” he points out. “I don’t think that [AI is] going to suddenly turn our brains to mush.” But staying sharp will take effort on our part.

When AI replaces the work, skills fade

A new wave of research is sketching out how our brains deal with help from AI. The answer can sometimes look grim.

Take one 2025 study of doctors who’d been trained to spot growths — called polyps — on patients’ colons. Doctors got an AI tool’s help spotting polyps for three months. Later, when the doctors returned to working alone, they didn’t detect polyps as well. This suggests the doctors “kind of forgot what to look for,” Cash says.

The same pattern plays out with other skills, such as learning a new math concept. One 2025 study looked at high-school students using an AI tool like ChatGPT to solve practice problems. The AI seemed to help. But take the tool away, and those students did worse than ones who never used it.

Something similar can happen with reading comprehension. Consider people who used an AI tool to solve practice SAT questions. Those people had a hard time coming up with correct answers when the AI was taken away, a new study finds. After losing their AI helper, people were also more likely to skip questions.

That was “particularly concerning,” the study’s authors conclude, because persistence is key for learning. These researchers shared the findings at the 2026 Conference on Language Modeling in San Francisco, Calif., on October 8.

All these results suggest people can struggle once they let AI take over their tasks.

For Cash, the lesson is “use it or lose it” — where “it” is your own brainpower. Don’t let AI do all your thinking for you, he says. “If you’re not engaging with the cognitive work, you’re not going to learn the skill.”

AI can help when it keeps people thinking

Using AI doesn’t have to mean your skills will crumble. You can put in the effort needed to learn even as you’re getting digital help, says Alp Sungu. The key is using AI tools that offer guidance or examples, rather than spitting out answers.

Sungu is an economist at the University of Pennsylvania in Philadelphia. He worked on the study of high-school students using AI for math. In that study, his team designed an AI tool with some limits. It worked like ChatGPT but didn’t give away answers. Instead, it offered hints and encouragement. That approach helped students learn as well as those who just studied with books.

Using AI as a tutor is different from using it as an answer machine, Sungu says. Students still had to put thought into their studying. The struggle of working through a new concept is where real learning happens, he says. “When you’re solving a problem, you need to get your hands dirty.”

For those willing to put in some effort, AI can be a good teacher, says Benjamin Lira Luttges. This behavioral scientist also works at the University of Pennsylvania.

An AI tool can help people learn how to edit cover letters, his research has shown. After practicing with either an AI tool or human feedback, people edited a poorly written cover letter without any help. Then, human judges picked which letters would likely land a job interview. Letters from both groups were equally likely to get picked, Lira Luttges’ group reports. The team shared these results on arxiv.org in February. (Studies posted there have not yet been vetted by other scientists.)

The AI tool in that study had taught by rewriting an entire cover letter itself. That may sound like its students weren’t doing much thinking. But it’s not what the team found. People didn’t just accept the AI-revised draft. They sat with it for a while, tinkering with the language. “They were still working,” Lira Luttges says.

This hints that AI can teach by example — “if you have people who are motivated to learn,” he says. In short, AI systems that keep people’s brains engaged might be able to sharpen some of our skills.

Still, we should be careful in choosing what thinking we outsource to computers, experts say. If there’s a skill you value, don’t give it over to AI, Cash advises. “We know skills come from practice and are maintained through practice.”

Lira Luttges agrees. People should think about what skills they want to protect, he says. That might be coming up with the perfect way to phrase a sentence. Or it may be devising a tasty dish using items in your pantry.

“You shouldn’t delegate the things that you enjoy and that give you meaning,” he says. “Outsource the crap, not the craft.”

If you have a dog, you likely know how good these animals are at reading subtle cues. Jangle a key or dangle a leash and many pooches know it’s time to go outside. But man’s best friend may be even more in touch with our feelings than we realize. Dogs can distinguish between certain emotions just by looking at our faces, a new study suggests.

The results offer a new peek at what’s going on in canine brains. That could offer clues to how dogs’ emotion-sensing skills evolved, says Laura Cuaya. She’s a neuroscientist at the University of Vienna in Austria. The findings also give another reason to smile at any pup that crosses your path, she adds. They just might understand.

Past studies had shown dogs can tell if someone’s face is wearing a positive or negative expression. The two emotions seem to trigger activity in different areas of a dog’s brain. But many questions remained. Could dogs tell the difference between two negative emotions, such as anger versus fear? And to read those emotions, did the faces need to be familiar to the dog?

To find out, Cuaya and her team started by training 14 family dogs to stay still inside an MRI scanner. “They are very, very good at the task,” Cuaya says. The machine would then track the dogs’ brain activity.

In one experiment, dogs looked at images of unfamiliar faces with happy or neutral expressions. Happy faces lit up a reward region in the animals’ brain more than neutral faces did. In another experiment, the researchers tested several negative emotions. The dogs’ brains reacted differently to angry or sad faces versus fearful ones.

The researchers shared these findings August 10 in iScience.

This is the first evidence that dogs can distinguish between negative emotions on our faces, Cuaya says. It suggests dogs can process emotional information in more detail than scientists once thought.

But the findings might not be the same in all dogs, Cuaya cautions. Her group focused on border collies and golden retrievers.

Her team is now working on another MRI study in dogs. This one looks at how the canine brain responds to a certain sound that dogs make. It’s thought to be their version of laughter.

What about cats? Cuaya’s team is not planning any MRI studies in them. Our feline friends may be a bit harder to coax into sitting still for brain scans.