Neurodivergent students call generative AI essential, but see risks

Students said AI can help manage academic demands while interface problems may disadvantage people with executive function differences.

Editorial illustration — not from the study.

Researchers observed 12 neurodivergent and 12 neurotypical university students completing an academic task with Microsoft Copilot. They then held focus groups with 14 participants to explore broader views of generative AI in academic work.

Both groups used the system strategically, focusing on completing tasks efficiently, evaluating and refining AI output, and maintaining academic integrity. Neurodivergent participants also described using AI to manage energy and cognitive processing, and said some interactions with the system felt more human. In their focus group, students called generative AI essential scaffolding for academic demands, while describing tensions involving authenticity and over-reliance.

How students used AI

The study identified three themes: approaching an academic task efficiently, evaluating and refining AI-generated output, and maintaining academic integrity. Both neurodivergent and neurotypical students showed strategic use of generative AI.

The neurodivergent students in this sample explicitly discussed cognitive processing needs and using AI to manage energy. They also described humanising some interactions with the system. In the neurodivergent focus group, students characterised generative AI as essential scaffolding for managing academic demands, but linked that support to concerns about authenticity and over-reliance. Interface design challenges seemed to particularly disadvantage users with executive function differences.

Why inclusive design matters

The findings point to a tension in AI-supported education: generative AI may provide important academic support for some neurodivergent students, while aspects of its design may make access harder or encourage dependence. As higher education institutions integrate these tools, the study argues that inclusive design and evidence about diverse student needs are needed to support equitable access.

Evidence and caveats

This was a sequential, multi-method qualitative study, not a test of academic performance or a measure of whether generative AI improves learning. Researchers observed 24 students—12 neurodivergent and 12 neurotypical—using Microsoft Copilot for an academic task, then conducted focus groups with 14 participants, split evenly between the two groups. The findings describe participants’ experiences and perceptions in this sample and may not represent all neurodivergent or university students. The abstract does not specify the university setting, the academic task or how participants were recruited. A neurodivergent research advisory board co-designed the procedures.

Programmable hardware speeds matrix calculations up to 6.3 times

Researchers tested a programmable hardware design that imitates selected quantum-computing ideas to run matrix calculations in parallel. In representative benchmarks, the design reached up to 6.3 times the speed and 4.1 times the energy efficiency of conventional matrix co-processors and embedded graphics processors.

2 days ago·13 views

AI & Computing

AI and docking agreed in some drug screens, but not all

Researchers evaluated three AI systems on hundreds of molecules bound to a SARS-CoV-2 protein and in three prospective drug screens. The results suggest that combining AI-based predictions with traditional docking may help prioritize candidates, although no single approach worked best in every test.

4 days ago·50 views

AI & Computing

A hardware monitor detected simulated attacks on car networks in 14 microseconds

Researchers developed a hardware-assisted system intended to detect message injection, spoofing, flooding and impersonation attacks on automotive networks. In tests using simulated traffic and a prototype automotive processor system, it reported an average detection time of 14 microseconds and less than 2 percent CPU overhead.

5 days ago·42 views