Logo

I went from AI illiterate to building an AI teaching assistant: here’s what I learned

See how a simple question can lead to innovative, AI-enabled teaching, such as an AI customer that develops students’ sales skills
Fabio Goncalves de Oliveira's avatar
17 Nov 2026
copy
  • Top of page
  • Main text
  • More on this topic
An AI chatbot floating above a man's hand
image credit: Sansert Sangsakawrat/Getty Images.

Created in partnership with

Logo

You may also like

We need to address the generative AI literacy gap in higher education
6 minute read

A few years ago, the most advanced technology I used regularly at work was a Microsoft Excel pivot table. Today, my students rehearse sales conversations with an artificial intelligence (AI) customer that I built for them.

The journey between those two points has changed how I think about AI literacy.

I want to share what I learned moving from being mostly AI illiterate to becoming someone who uses, develops and critically evaluates AI tools in my teaching. My message is simple: academics do not need to become AI engineers but we do need to understand AI well enough to decide where it adds value, where it does not, and how we remain in control of its use.

Until 2018, I was a sales director in a multinational company. I understood sales strategy, customers and teams but very little about AI. When I moved into academia and began a PhD in business and management, my research brought me closer to the relationship between digital technologies and decision-making. AI gradually moved from something I read about to something I wanted to understand and use.

Reading about AI gave me knowledge. Using it to solve a real teaching problem forced me to develop AI literacy. 

Much of the conversation about Generative AI in higher education focuses on what it can do for students. It can summarise readings, generate ideas, write text, analyse information and answer questions.

My experience has led me to ask a different question: what can AI enable students to do for themselves?

I came to this question while teaching sales and business development, where students need to learn how to conduct a sales conversation. They can study sales models and discuss them in class but developing the skill requires practice. Students need to ask questions, listen, respond to a customer, handle objections and make decisions in real time. 

Practice, however, creates a problem. Students need someone willing to repeatedly play the customer, at a convenient time, without becoming bored or changing the nature of the exercise. 

I saw an opportunity for AI. So, I built one.

I developed Debbie, an AI customer that students can speak to before completing their assessed sales role play. Debbie plays the customer while the student plays the salesperson. The conversation follows the stages students have learned in class, but the AI responds to what the student says. At the end, it provides feedback aligned with the skills they are expected to demonstrate.

Students can practise again. And again. They can make mistakes without embarrassment and return when they feel ready. 

Developing Debbie forced me to learn far more than how to write a good ChatGPT prompt. I had to understand prompting, retrieval-augmented generation (a technique that allows to large language models and generative AI tools to ground their answers based on proprietary data), data protection, cloud services and how different AI components work together. More importantly, I had to keep asking a pedagogical question: What should the AI do, and what should remain the responsibility of the educator?

That question has become central to my approach to AI.

The purpose was never to replace me as an educator. I designed the system to extend an activity that is difficult for me to provide individually and repeatedly to every student. I retain responsibility for the learning outcomes, activity structure, assessment criteria and how the technology fits into the module.

There was another benefit I had not fully anticipated. Students received individual opportunities to practise, while I gained a clearer picture of where they were struggling. Their interactions with the system could inform what I needed to revisit in my teaching.

Students also valued being able to practice when and where they wanted, repeat the experience and receive immediate feedback. For them, AI was not producing an assignment. It was creating an opportunity to practice the skills needed to complete it themselves.

This experience changed my understanding of AI literacy. It is not simply knowing how to use ChatGPT, nor does it require every academic to become a programmer. It means understanding enough to make informed choices about when AI should be used, what role it should perform and where human judgment must remain.

My advice to colleagues who feel behind with AI is therefore not to start by trying to learn everything about it. Start with one problem in your own practice.

Ask: What am I trying to help my students do that is currently difficult to provide at scale? Could AI support that activity without doing the learning for them? What do I need to understand to remain in control of the process?

Then, learn what you need to solve that problem.

That is how I developed my AI literacy. I did not learn AI first and then search for somewhere to use it. I had a teaching problem, experimented with AI and learned what I needed along the way.

AI has become a vector for continuous learning in my professional life. But the most important lesson has not been how powerful the technology is. It has been understanding that using AI purposefully requires us to decide what we want it to do for us, and, equally importantly, what we should continue doing ourselves.

Fabio Goncalves de Oliveira is lecturer in entrepreneurship and innovation and programme director for undergraduate business and management (entrepreneurship and HR&OB) at Henley Business School, the University of Reading.

If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.

You may also like

sticky sign up

Register for free

and unlock a host of features on the THE site