AI Inequality in Higher Education
Team name : AI Inequality in Higher Education
Use of AI tools :
Claude -> For structuring and refining ideas and reformulating written contributions. Claude help me with the language.
HeyGen -> To create the video because I am not a presenter. But also because it illustrates something essential : today, you can do almost anything with AI. And that is exactly why students use it.
How did these tools support your work ?
Claude helped us organize our thinking across multiple phases. It helped us maintain a consistent academic tone across the contribution. Importantly, the core ideas, including the inclusive AI comparison platform, the NGO relay model, and the progressive curriculum integration, came from our own reflection. AI was used as a thinking and drafting tool, not as a replacement for our own analysis.
HeyGen : I provided my ideas, my pitch, and chose the voice and the avatar. Then HeyGen did its job, and now you have a presentation that catches your eye.
Initial contribution : How could you resolve AI inequality in higher education ?
External feedback & contributions :
Peer comments pushed us to prioritize our solutions by timescale, focus on professor training as a first step, ground our argument in real student testimonials, and redesign our peer-educator model to avoid reproducing the inequalities we aimed to solve.
Final contribution :
Several studies conducted in universities illustrate how deeply AI has become embedded in students' daily lives. Students are aware of the risks, particularly the loss of autonomous thinking and critical reasoning. Moreover, despite the fact that AI is widely used in academic work, even when not permitted, inequalities remain significant: between individual students, between universities, and between countries. If we acknowledge that AI must be integrated into education, we must do so without deepening existing inequalities.
Our initial contribution presented many relevant ideas but lacked a clear center of gravity. We therefore chose to focus on what is most immediately actionable: responsible education in how to use and prompt generative AI tools such as ChatGPT or Claude. We propose introducing this education progressively, first as an optional course within higher education curricula, allowing students to learn how to write effective prompts and use AI critically, then integrating it fully into standard curricula as awareness and institutional readiness grow. This approach respects the fact that AI literacy, while increasingly essential, is still unevenly understood across institutions.
Two structural barriers complicate this: the language barrier and the financial barrier. Technical efforts to adapt AI models to low-resource languages such as Swahili or Wolof already exist, as do some initiatives addressing cost. However, structured educational frameworks teaching students and professors how to use these tools responsibly — rooted in local cultures and relayed through NGOs already trusted by local communities, do not yet exist at scale. This is where our contribution begins.
Our final and most original contribution is the creation of an inclusive AI comparison platform. This ecosystem brings together all available AI tools in one searchable environment. Users enter their needs, their language, and their budget, and the platform returns a ranked list of tools matching their criteria, with filters for price, language support, and effectiveness. Rankings are established by local experts and NGO partners who assess each tool's effectiveness in specific languages and contexts. Users can also submit feedback, but reviews are not published automatically, each submission must include the prompt used, the result obtained, and the reason for dissatisfaction. This moderation process ensures review quality while simultaneously teaching users what an effective prompt looks like.
The platform's reach can grow organically through education itself: professors leading AI literacy courses would naturally introduce it to their students as a practical tool, creating a diffusion channel that requires no marketing budget or institutional mandate. As the platform gains visibility, it also creates competitive pressure that could incentivize AI providers to introduce student pricing plans where none currently exist, further reducing the financial barrier that no single solution can fully eliminate.
The goal is simple: make all available options visible, so that finding the right tool becomes easier for everyone, regardless of language, budget, or geography, and ensure that AI in higher education becomes a force for reducing inequality, not amplifying it.
Please see the full written contribution above, including introduction, problem statement, education framework, platform proposal, limits, discussion, and conclusion : https://drive.google.com/file/d/15-tjpFDIoRV26M_RXacgi5U7Mx--GUEu/view
Reflection on the process :
How did your contribution evolve during Phase 2 ?
Our initial contribution was broad and ambitious, it covered many solutions across multiple actors (universities, governments, NGOs, civil society) without a clear center of gravity. While the ideas were relevant, the proposal lacked focus and a realistic sense of what could actually be implemented and by whom.
Phase 2 fundamentally changed this. Through peer feedback and internal discussion, we progressively narrowed our proposal around a single coherent logic : responsible AI education as the most immediately actionable response to AI inequality, supported by an original and concrete tool : an inclusive AI comparison platform.
What feedback influenced your work ?
The first comment pushed us to think in terms of timescales rather than flat lists of recommendations. The second led us to prioritize professor training as the entry point for university-level action. The third, reinforced by Caroline Beslin's keynote, made us integrate student testimonials and real survey data into our introduction, grounding our argument in lived experience rather than theory alone. The fourth, from Johanna of the T-Twice team, was the most intellectually challenging: she pointed out that our own solutions could reproduce the inequalities we were trying to solve. This forced us to think carefully about the design of our peer-educator network, the role of NGOs as cultural relays, and the moderation system of our platform.
What changes did you make as a result ?
One of the most significant changes in our process was the decision to validate our hypotheses with real data. We actively searched for surveys and academic studies that could ground our claims in evidence. Reading other teams' contributions also played a role: seeing how other groups approached similar problems helped us identify what was missing in our own proposal and pushed us to go further. This research process, combined with peer feedback, is what ultimately led us to our most original contribution, the inclusive AI comparison platform. It emerged not from a initial plan, but from a progressive deepening of our analysis, step by step.
How did this process strengthen your final proposal ?
The iterative feedback process forced us to be more honest, more precise, and more realistic. We moved from a wide-ranging list of recommendations to a focused, sequenced, and actionable framework with a genuinely original contribution at its center. The process also taught us to acknowledge the limits of our own proposals openly, which we believe makes the final contribution more credible and intellectually rigorous.
Video : https://drive.google.com/file/d/1DsJQvjh5Xs21FmQpkT4fJqsLcgNAgMeE/view
Share