Artificial Intelligence (AI) tools now shape how many students tackle tasks such as essay writing, problem-solving, and even brainstorming ideas. Across online platforms, users claim that their reliance on AI has compromised their vocabulary, writing abilities, and creativity, raising concerns about a weakening of cognitive skills overall. However, from a scientific perspective, does sustained use of external cognitive tools truly alter how these abilities are exercised or maintained? 

In a written exchange with The Tribune,  Nandini Asavari Bharadwaj, a PhD candidate in McGill’s Department of Educational & Counselling Psychology, explained how these external tools change people’s engagement with and maintenance of their cognitive skills.

“Generative AI tools are relatively new, and more research is needed to understand their long-term impacts on human cognition; however, current research in the area suggests generative AI use may have various impacts on learning and cognitive processes like memory and attention,” Bharadwaj wrote. 

Bharadwaj also emphasized that any tool we use will shape how we think and process information.

“Generally speaking, the use of any external tool during cognitive processes can impact how we think and perform cognitively, from using a simple pen and notepad to computers, and now to generative AI,” she explained. “Moreover, tools are not entirely neutral by themselves. Our use of these tools is driven by what the tools can offer as functionalities, what are known as real affordances of tools and our perceptions of what they can do as well, known as perceived affordances.”

Bharadwaj added that researchers have found how AI tools can both limit and enhance cognitive skills, depending on their use.

“On the one hand, researchers have raised serious concerns that these tools, if used excessively, especially by novices, could reduce opportunities to develop and practice skills or ‘productively struggle,’” Bhardawaj wrote. “On the other hand, research has shown that generative AI can offer exciting new ways to engage with educational content, augment existing knowledge, and boost students’ reasoning and analytical skills.”

As such, one may wonder whether to use AI tools at the start, middle, or end of a task, or perhaps only for a portion of the task, to mitigate a potential weakening of their cognitive skills; however, researchers remain unsure. 

“Some researchers have recommended using AI systems for brainstorming, review, or editing activities that still place the onus of creation on the student. Others recommend it to be used as a tutor or guide, rather than a full teacher, to promote mastery in learners,” Bharadwaj wrote. “Ultimately, in education, we want students to develop thinking and writing skills, not just hand in assignments. Hence, applications of any technology for learning, including AI, should always be guided by pedagogical and learning goals.”  

Looking ahead, these elements underscore the need for thoughtful interaction with AI tools, which requires understanding how they work and why they are being used. 

“Students should reflect on what skills they are trying to develop, where they might need more or less teacher intervention, and what their overarching learning goals are. This will encourage more thoughtful use of AI tools for skill development in the long term as well,” Bharadwaj wrote.

Bharadwaj also stressed the importance of developing AI literacy skills.

“Students must understand AI systems and their capabilities, i.e., develop their AI literacy skills, before any extensive use of them. By doing so, students can ensure they understand how these tools work and are informed on issues like system hallucinations, safety, privacy, alignment, and transparency. This can also promote best practices for working with AI systems.”

So, fact or fiction: Can AI reduce cognitive skills over time? Limited research offers mixed findings, but the answer ultimately depends on how individuals choose to use AI tools.

“It is not just a simple question of whether AI tools are good or bad for cognition. Rather, we need to get into the details of how they are being used and how the learner is engaged in the learning process,” Bharadwaj wrote.