An alarming number of Kenyan university students have increasingly turned to artificial intelligence (AI) tools in place of traditional studying, raising concerns about academic integrity and future employability in a competitive global job market, according to a new study.

Almost all Kenyan students now use various AI tools for their daily academic work, with nearly a third admitting that these chatbots have “somewhat replaced” studying for them, a survey by Nairobi-based research firm Stadi Analytics reveals.

The research indicates that 90 percent of students in higher learning institutions engage with AI regularly, with a majority – 35 percent – using these tools several times each week.

Conventional chatbots such as ChatGPT, Co-Pilot, Gemini, and Perplexity, which draw information from multiple online sources to answer practically any academic question, are the most popular among students.

They also rely on grammar and paraphrasing tools, coding assistants, summarisation software, and translation applications, with at least a quarter of respondents reporting frequent use.

Students cite multiple reasons for turning to AI, including to understand complex concepts, save time, improve assignment quality, generate creative ideas, and proofread work.

Charles Muriithi, a third-year Bachelor of Arts in Linguistics and Literature student at Chuka University, confirms that AI has transformed how he learns.

“AI has improved my learning process. It makes my work so much easier and saves me time when doing assignments. I can recommend that all students use it, honestly,” Muriithi told _The EastAfrican_ in an interview.

Stadi Analytics’ research found that 84 percent of students using AI indeed reported marked improvements in their academic performance.

The study also revealed that AI use spans all disciplines and levels of study but is most prevalent among final-year students and in courses such as business, economics, medicine, law, social sciences, and arts and humanities.

Use was least common in agriculture, environmental studies, education, and engineering – fields considered more hands-on than theoretical.

Yet, while AI use has clear benefits, it is the growing dependency on AI tools that worries, according to Elias Muhatia, the lead researcher of the study dubbed AI in Higher Education: Comprehensive Analysis Report, published last week.

The study found that about 29.5 percent of students using AI admitted feeling uneasy about their overreliance on these tools instead of engaging in traditional studying to understand concepts.

“Even more concerning is that over 10 percent of these students report being unable to remember content they submit as assignments because AI handles everything,” Muhatia said.

The research further revealed that more than 42 percent of AI-using students largely agree that the technology significantly reduces critical thinking, demonstrating a clear awareness of the cognitive risks associated with AI use in academic work, he added.

Muriithi emphasises that the impact of AI depends on how students use it and is not uniform across the board.

“It is true that AI can enable lazy students who simply copy and paste, but it can also support students who use it only for inspiration or guidance, not to do the job for them. So, I think the choice is really up to the student,” he said.

Despite the widespread adoption, the study found that faculty are often unaware and non-complicit in students’ use of artificial intelligence, and most institutions have yet to establish clear policies to guide responsible usage.

Muhatia argues that safeguarding the integrity of the Kenyan education system in an era of widespread AI adoption requires a shift in focus.

“With near-universal student adoption and sophisticated risk awareness, the challenge shifts from preventing AI use to optimising AI integration for educational enhancement,” he said.

Muhatia underscores the need for policies and strategies that harness AI’s benefits while mitigating its risks.

AI is a broad field of computer science focused on building systems capable of performing tasks that typically require human intelligence.