Threat or Opportunity? AI Takes Academia by Storm

The artificial intelligence chatbot that has stormed into our lives poses new challenges to teachers and lecturers. But with a bit of creative thinking – and perhaps help from the bot itself – this serves as a major opportunity to deepen learning, acquire skills relevant to the industry, and even tell when the person or computer in front of me is “jabbering with confidence”. How to do it? Read on.

In November 2022, the world changed all at once with the introduction of generative AI. It is hard to think of a field that ChatGPT and similar AIs have not profoundly altered or will alter. Yet, when it comes to scholastics and higher learning in particular, the change is unusual, extreme, and immediate. How is academia to take this revolution? Should it see it as a threat or an opportunity? Before we try to answer this, it is important to understand: the revolution is already here and moving forward. Despite calls to halt development, block access from campuses, or as some voices in Italy call for, “block access entirely”, the AI train has left the station.

“We’re past the point of no return,” says Dr. Sharon Yalov-Handzel, a senior lecturer at Afeka College’s School of Data Science and Artificial Intelligence. “There is no point in blocking an effective tool that is at our disposal anyway. It’s unnecessary and unrealistic. I advocate the opposite approach: if there is a technological tool – by all means, make use of it. It is precisely through this tool that you can acquire and train for proficiencies that today’s industry requires, such as critical thinking and telling the essential from the inessential. Because at the end of the day, as one student told me, the chatbot ‘jabbers with confidence’, in a way that’s much harder to spot than human ‘jabber’. And this is precisely the test.”

The End of Summarizing

As an expert in Natural Language Processing (NLP), one of the leading fields in AI today, Dr. Yalov-Handzel is very familiar with the thing that is currently changing our lives. She says that she herself spends a lot of time on ChatGPT, and often uses it in her work. “You can’t rely on it a hundred percent, you still need to be critical about the information it provides, but it certainly saves hours of work.”

She is aware of the potential costs of this new world, the potential for losing critical learning skills due to the dramatic technological shifts. On the other hand, Yalov-Handzel sees the opportunity which these new tools provide for teaching and pedagogy, for practice, feedback, and personalized learning – if only we use them wisely. “Nowadays, it seems that the ability to evaluate scope and size matters more than long division; that the scientific thinking used to reach a solution matters more than the end result.”

Yalov-Handzel cites the standard pedagogic approach that divides learning and conclusion-drawing into five levels, from basic to complex. “Academia,” she says, “used to focus on the mid-levels with the occasional nod upwards. Students spent a lot of time on ‘recap’, on sourcing materials and writing summaries. Now AI does the work for them. Rather than lamenting this, we should take it to the next level. We in engineering and exact sciences have a lot to add, to scale heights and plumb depths that are beyond the reach of existing AI models: creative thinking and profound understanding which aren’t necessarily strictly linguistic. Even assuming that they’ll become more precise (even in Hebrew), the models’ ability to attain high cognitive levels of complex problem-solving is still far off,  since we don’t know how to program consciousness and put in into a computer. We’ve got a good few years until that happens.”

Question in Three Prompts

We asked Dr. Yalov-Handzel for a practical example. “The starting assumption is that students already possess the basics, so I demand a higher level of understanding, analysis, and application. In the programming languages course, I illustrate principles using Python. The exercises I used to hand out before this semester were considered complex and required quite a bit of thinking, but now the chatbot can solve this type of exercise pretty well, even if not with total accuracy. So I’ve decided to present a problem that is harder by three orders of complexity. The kind of problem that, if you cleverly break it down into three prompts (the queries fed into the AI), the chatbot can solve the problem. But the combination requires additional thinking, understanding, and sophistication.”

Another example is a new NLP course, which Yalov-Handzel has had to redraft even before she could teach it for the first time. “The algorithms I originally planned to teach are small fry for ChatGPT.” Yet what of those who aren’t versed in the new tools yet? Is this another instance of technology widening gaps? “I assume that the use of this tool is already a given, certainly among future engineers who are mostly young and technology-oriented. But there’s room in this course for teaching how to be an effective prompter. Perhaps not for credit but to catch up. This technology has the potential to close gaps, because it enables ‘individualized learning’ for everyone.”

We now return to our opening question: Threat or opportunity? Because academia, like all of us, is learning as it goes. “The whole world is in shock, trying to process and understand. Regulators are on it, the OECD is on it, the Biden administration has issued a call for anyone working in AI to give their two cents on regulation and ethics in order to form a strategy. It changes thinking patterns, calls for a sociological and ethical response, and requires lecturers and educators at all levels to adapt their curricula. There’s been a technological leap, and it’s hard to catch up all at once – this new technology will tap into quantum computing, and we’ll witness another leap – but we will remain relevant to the changing reality, and especially to the needs of the industry.”