John Bicknell, Co-Founder and Chief Product Officer of Tutello, has spent most of his career building educational tools. Based in Gran Canaria, he works alongside Krystian Paszkiewicz, Tutello's Chief Technology Officer, and leads the development of the platform. He believes that those institutions who embrace AI and lean into it at this early stage will have a significant advantage over those that push back against it. Covering the topics of LLMs, trust, educating the educators, and understanding AI as a tool rather than a magic wand, we sat down to dissect the current landscape of AI in education.
“Anyone who says they know where AI is going shouldn’t be trusted!” This is a phrase that Tutello’s Co-Founder and Academic Lead, Professor David Lefevre of Imperial College Business School in London, often repeats to the team. While no one can be certain where AI is heading, what we do know is that we’re at the very early stages of a new era of AI in education.
And so, before discussing the challenges institutions face in engaging with AI, I wanted to begin by asking John to take a step back and reflect on the recent boom of Large Language Models. “They went from being an interesting thing to play with to suddenly having real value,” he said. “They probably got overhyped at one point, but I think it will settle back down now. We go through that ‘hype’ cycle of expecting too much, and then people almost discount it again. But I think the big important step was its transition from being something that did quite cool things from time to time to being a really useful tool.” American scientist, Roy Amara, considered this phenomenon, which is now known as Amara’s Law (or, as some refer to it, Bill Gates’ Law): “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.”
“I think it will take some time for us, for everyone, to figure out how it best fits in with software and how we work", John says. "I think that process will take a few more years. There’s an expectation that everything will change overnight, but it will take a long time to determine the fine-tunings and adjustments needed to make it fit into your workflow and for it to start working for you.”
To give some insight, the institutions Tutello is working with are all using the platform in slightly different ways. Additionally, some institutions are uploading administrative content, such as deadlines for assignments, alongside standard course materials. And so, over time, our clients will learn where AI works best for them, and there’s no doubt that what works for one faculty may not work for another.
Large Language Models (LLMs) have recently made headlines due to concerns about their financial implications and ethical considerations, particularly regarding data privacy and misinformation. Given these complexities, I asked John why Tutello isn’t tied to a single LLM. “It’s all about options, really,” John explained. “We currently use OpenAI [as our base LLM that sits underneath Tutello] because, at the beginning, they were hugely more capable than the competition and were reliable and consistent. Since then, the competition has caught up and, in some cases, even overtaken them. With the recent launch of Meta’s open-source model, Llama 3.1, we now have an open-source option as well as commercial offerings. This allows us to fit into whatever our clients prefer, giving them more control over their choices. You might have a client who is extremely keen on privacy and open-source solutions and doesn’t want to work with big corporations, or you might have a client that has a deal with OpenAI.”
"AI is a tool people must learn to use, not a magic wand"
When we have introduction calls with institutions, what we’re really looking for is that willingness to learn about AI’s potential and applications. John thinks that the biggest challenge we face in integrating AI into existing workflows is “helping people understand that AI is a tool they must learn to use, rather than a magic spell. There are times when AI does perform like magic. But, while there are instances where it excels, those are few and far between.”
And so, while AI has made it easier than ever to learn new concepts at our fingertips, we cannot forget to think critically, just as if you were reviewing information from elsewhere. John often compares using AI with searching on Google, “if you pick the first answer every time, you can end up at some rubbish site. But if you learn how to rephrase what you type into Google and learn some of the more advanced tools, you’ll have a much better chance of getting some useful stuff. And then you figure it out from there.”
Asking John about the rate of development and familiarisation of AI in Higher Education, he feels that “it's going to be a 5 – 10-year deal to actually get to that point where AI becomes the default rather than something that’s a novelty, new and exciting.”
“Almost frustratingly, we've got to start where everybody is now and gently feed it into what they're already doing,” John reflects. “From there, we can then evolve until what we can see will be the end game. I’d say we’ve not nearly scraped the surface yet. The potential is massive, but the main technique [right now] is understanding how a particular job is being done, identifying how we can add a little bit of AI into it and moving forward from there.”
"We've got to start where everybody is now"
It’s important to highlight that institutions embracing this new era of AI in education are not making monumental changes to their curricula overnight. Some of Tutello’s partners have chosen to integrate AI across multiple courses, while others prefer to start with just two or three, with plans to expand later. Given the significance and power of AI, it’s essential to approach its integration in a controlled manner.
At this stage, then, engaging and training faculty is key. They need the knowledge and skills to interact effectively with AI, as they will be the ones endorsing a chosen product to students.
With this "lean in and learn" mindset, Tutello encourages institutions worldwide to begin exploring how tailored, controlled AI solutions can enhance workflows and transform learning outcomes for both students and faculty. Institutions that take the first steps now, rather than waiting until AI’s advantages become too substantial to ignore, will be the ones to reap the most significant rewards.
