Skip to content
ashley cheng

Nurturing human agency in the age of AI

Visit AI Consensus (opens in a new tab)

Bringing people together.

Students@AI Seoul

2025

I designed the hackathon’s visual identity in Photoshop and Illustrator, from the badges hackers wore to the prize boards at the closing ceremony.

The event poster Select a piece to bring it forward

Students@AI Buenos Aires

2025–2026

I designed posters and slides for our monthly events with local universities, along with the AIC Labs grant poster.

AIC Labs · Buenos Aires grant program Select a piece to bring it forward

Students@AI Taipei

2024

I invited the speakers, coordinated partnerships and logistics, and co-hosted a conversation about AI and learning in Taiwan.

Guest speaker panel Select a piece to bring it forward

From campus research to a global conversation.

2025

The Minerva AI report.

We surveyed 50+ students across four class years and led a discussion on AI and learning with students representing all five colleges. The AI Night report drew on that discussion and its accompanying survey to make recommendations for course and assessment design. It sparked debate among professors, and some put our suggestions into practice—giving students a voice in how the university responds to generative AI.

Ongoing

AIC Voice.

Through AIC Voice, we’re extending our impact globally, exploring how young people envision the future of education, their own futures, and the world they want to build.

We’re looking both across countries and within them to understand shared hopes alongside local experiences and priorities. Our goal is to bring those perspectives into decisions about AI and education, giving young people a meaningful say in what comes next.

Explore AIC Voice (opens in a new tab)

AIC in the press

AIC has worked with

  • DeepMind
  • Amazon Web Services
  • Elice
  • Perplexity
  • Samsung
  • Upstage
  • Wolfram Alpha

Minerva AI report · Selected findings

How students use AI, and what they want from their education

Authored by Ashley Cheng with the help of ChatGPT and Granola.ai. Discussion organized and led by Brady Rutherford. Convened by AI Consensus.

Where these findings come from

Our wider survey reached 50+ students across four class years—about one in ten students at a university of roughly 500. This excerpt draws on a smaller part of that work: an anonymous survey with 16 responses and a discussion with 17 students in December 2025. The discussion, called AI Night, brought together students from across the university’s five academic divisions. Most of the survey respondents also took part in the discussion.

Participation in the wider survey was not random. These findings describe the smaller AI Night group and helped us develop recommendations; they do not tell us how common these experiences are across the university.

AI was already part of studying

In the smaller survey, 15 of the 16 respondents used AI for the reading and exercises assigned before class. Five said they had used it to do that work entirely for them in at least one course. We wanted to understand what was behind those choices: when did AI help students learn, and when did it take the place of learning?

Pressure, purpose, proficiency

As students talked about their experiences, three themes kept coming up. We called them pressure, purpose, and proficiency.

Pressure: having enough time to learn

Students often began with how much they had to get through. One participant said they would choose “one extra hour of sleep” over doing better on the work due before class. Under that pressure, AI offered a way to finish, even when it meant doing less of the thinking themselves.

Purpose: seeing where the work leads

Students questioned the value of assignments that seemed to disappear as soon as a professor graded them. They wanted their effort to build toward something they could present, publish, or use beyond the classroom. That wish came through in one participant’s description:

“We should stake ourselves in the world… not just a paper that professor reads and goes in a digital bin.”
— AI Night participant

Proficiency: the quality of teaching

By proficiency, we mean how professors teach. Students described being more willing to explain their thinking and be open about AI use when professors asked thoughtful follow-up questions, answered questions in depth, and gave useful feedback. They wanted to feel that their understanding would be noticed and fairly assessed.

These themes were connected. An overloaded student had less reason to persist with work that felt disposable. Clearer goals and good teaching could help, but students also needed a manageable workload.

What could change in the classroom

We used what students told us to propose changes to assignments, teaching, and grading. These were recommendations for the university to try; this study did not test their effects.

  1. Give the course a destination students can see. Show students what they will build or solve at the start of a course, then make readings and exercises contribute to that project. Shorter tasks, graded mainly for completion and feedback, could give students room to practice without needing every answer to look perfect.
  2. Let students propose work they care about. Give them the option to apply course ideas to a problem of their choosing, with their professor’s approval. Ask them to explain why they chose an approach, what alternatives they considered, and what evidence supports their decisions.
  3. Give that work an audience. A project could involve gathering feedback from people who use a local park, working with a community organization, or publishing something others can read. Building those conversations into an assignment gives students a reason to care about what happens after it is graded.
  4. Make more room for questions and feedback. Support professors in following a student’s idea, asking another question, and adjusting the discussion. Give students useful feedback while they can still act on it, and ask how the course is going early enough to make changes.
  5. Ask students to explain how they used AI. Use citations and short reflections within assignments to show what AI contributed and how students judged its suggestions. The report recommends recognizing thoughtful, open use without requiring extensive screenshots or chat logs that students described as extra work.
“When students see that early work directly feeds into something they will later present, publish, or defend, preparation becomes an investment rather than a hurdle.”
— From the report’s recommendations on preparing for class

Adapted for this portfolio from Student sentiment, AI use patterns, and a design agenda for AI-native learning, the AI Night report. Findings, the three themes, and recommendations draw on sections 1–4 and 7–8; this is a summary of the report.