Because of artificial intelligence’s rapid advancements, people can now use free generative AI tools to solve their homework, write code, finalize projects and even plan out whole vacations. These tools can give us almost instantaneous answers that save us time. But offloading our thinking to this kind of advanced computer can have costs ― not just because the answers can often be incorrect but also because it can have unforeseen consequences to how we reason through and recall ideas.
How we critically think and problem-solve on our own is being changed by AI tools in surprising ways, as one new study led by Microsoft and Carnegie Mellon University researchers found.
Researchers surveyed 319 workers in mostly computer, math, arts and design fields who use generative AI tools such as ChatGPT at their job. Study participants shared 936 self-reported examples of how they used these AI tools, as well as their perceived efforts to think critically while using them.
The data showed that people who trusted generative AI’s ability to complete certain tasks had more answers associated with “less critical thinking effort,” according to the study.
“While GenAI can improve worker efficiency, it can inhibit critical engagement with work and can potentially lead to long-term over-reliance on the tool and diminished skill for independent problem-solving,” the researchers concluded in the study.
Of course, the stakes mattered too. When the job stakes were higher for getting the right answer, workers said they evaluated the generative AI’s results much more thoroughly and cross-referenced them with other sources.
And not everyone trusted an AI tool’s answers. Workers who rated themselves as more confident in their own skills said they reviewed outputs more critically. Those who were less experienced at a task, however, said they relied more heavily on AI-generated answers.
“Reviewing outcomes is always important, but it’s especially critical for those who lack deep expertise in the topic they’re using AI to assist with,” Lev Tankelevitch, one of the Microsoft researchers for the study, told HuffPost.
Can you use ChatGPT and Gen AI tools without losing your ability to think critically?
The Microsoft study aligns with a growing body of research on how generative AI tools are changing the way we learn ― sometimes for the worse.
One 2024 study in Turkish schools found that students who used ChatGPT as a tutor to practice math problems did worse in math tests than those who did not, with researchers concluding that the students were using the AI tool as a learning “crutch.”
In Tankelevitch’s view, AI works best as a “thought partner” to complement humans. He cited a different study that found that students who had tutors supported by AI were more likely to master topics.
“The key: It was teacher-led. Educators guided the prompts and provided context, showing how a collaboration between humans and AI can drive real learning outcomes,” he said.
It’s important to note that Microsoft has generative AI services, therefore may have a more vested interest in encouraging people to not forego AI tools altogether. The study also only looked at a small population, so it’s hard to definitively prove that the same results would occur in a wider group.
Overall, this is not the first time we as humans have outsourced our thinking to new technologies to save time. Think of how we used to memorize phone numbers and maps with just our brains to get around, and now because of tools like Google Maps and phone contacts we do not.
But AI is much more advanced than a tool for phone numbers and maps. The big difference now with AI is that we are not offloading information but our thinking, said artificial intelligence researcher Michael Gerlich.
In his recent study, he found that the more people relied on AI tools like ChatGPT, the lower their critical thinking skills were. “And this is where what all those studies [like Microsoft’s] currently come to the same solution is when we offload the thinking process, then it has an impact on reduction of critical thinking,” he said.
So to put it bluntly: Is using generative AI making us dumber?
“If we just use it to offload our thinking, then yes,” Gerlich said. Ideally, people should use these tools to get proof for their hypothesis and ask ChatGPT for opposing opinions to avoid confirmation biases, he said.
If you want to be better at using AI tools, remember that you are the one doing the thinking, not the computer. “Those tools are not really thinking. They are predicting words based on trained materials,” Gerlich said. “I truly believe any student, any high school student, is smarter than GenAI in its current stage.“
In Gerlich’s research, he has found that people who are most at risk of losing their critical thinking skills to generative AI are people who are less educated, young people and people who are not confident with their own knowledge.
But no one is immune. In his study, “even top managers said it is very tempting to use [AI tools] because we are all so much under time pressure, and it gives [us] a quick solution,” he said.
The good news is that if reading all this research makes you worried about whether ChatGPT is turning your brain to mush, that’s a sign of being a good critical thinker. Take what one of the respondents in Gerlich’s study said about their growing reliance on AI.
“I find myself using AI tools for almost everything — whether it’s finding a restaurant or making a quick decision at work,” this middle-aged respondent in Gerlich’s study stated. “It saves time, but I do wonder if I’m losing my ability to think things through as thoroughly as I used to.”
For Gerlich, this kind of doubt is a positive sign that you are still doing measured skeptical analysis and critical reasoning that is necessary for being a strong thinker.
“If you ignore the risk, then I think it’s too late,” he said.