GenAI & Education: Enhancement, not Replacement

How to co-exist in the age of automation

Today, I’ll be joining colleagues from OpenAI, Google, Microsoft, Stanford, Harvard and other others at the first meeting of the GenAI Summit. Our shared goal is to help to educate universities & schools in Europe about the impact of Generative AI on their work.

One the questions we’ll explore at the summit is: how can we effectively communicate to education professionals that generative AI will enhance their work rather than replace them?

Here’s my TL;DR:

Educators are right to be unsettled

A recent study by academics at Princeton, the University of Pennsylvania and New York University found that out of the top 25 roles which will be most impacted by AI, 18 (72%) are educator roles.

The researchers used a benchmark known as the AI Occupational Exposure to evaluate how much services like ChatGPT could disrupt different professions. They concluded that within the education sector, post-secondary teachers of languages and literature, history, law, philosophy, religion, sociology, political science and psychology would be most affected.

But, disruption takes time (and isn’t inevitable)

AI has been around for some time. One of the things we’ve learned from the history of innovation is that change takes time (and isn’t inevitable).

Take the introduction of the printing press in the sixteenth century. At least for the first 200 years, it arguably did more to spread and firm-up existing ways of thing and behaving, rather than changing them.

GenAI - the 21st-century equivalent of the 16th-century printing press?

Similarly, the myriad of education technologies powered by generative AI that have appeared in the last six months or so have been built to streamline and accelerate what we, as educators, already do: faster grading, faster content, faster assessments.

Also, disruption does not = displacement

Georgia Tech University have been building a growing suite of AI-powered teaching tools for a decade now. Jill Watson and Agent Smith are AI-agents capable of answering student questions, hosting office hours and even connecting students who have similar interests to build social connections.

What we’ve seen at Georgia Tech is not a decline in the employment of human teaching staff but instead the liberation of staff to focus on where they bring most value: facilitating discussions, encouraging critical thinking, and fostering a collaborative learning environment.

We see similar patterns in other professions.

In medicine, AI has helped doctors to automate parts of their process which are repetitive and time consuming. It’s also helped doctors to become more effective in their roles. This has led many to ask the question, “Will AI replace doctors?”. There are many reasons that this would be challenging, including non-linear working methods (doctors often use creative problem solving, which is hard to programme) and the need for professional competency (AI solutions still require expertise to function effectively).

This latter point was made very powerfully in a recent example from the legal profession. While ChatGPT can assist lawyers by creating summaries of case notes and relevant laws and statutes and drafting documents such as contracts or agreements. This lesson was learned the hard way by lawyer Steven Schwartz, who used ChatGPT to generate case examples which transpired to be entirely hallucinated.

For now at least, ChatGPT is not a direct replacement for human expertise.

Embrace ChatGPT as a PA, not a threat

A recent controlled study found that ChatGPT can help professionals increase their efficiency in routine tasks by ~35%. If we keep in mind that the productivity gains brought by the steam engine in the nineteenth century was ~25%, this is huge.

As educators, we should embrace the power of ChatGPT to automate the repetitive tasks which we’ve been distracted by for decades. Lesson planning, content creation, assessment design, grading and feedback - generative AI can help us to do all of these things faster than ever before, freeing us up to focus on where we bring most value for our students.

BUT, as the recent case of Steven Schwartz reminds us - these tools need you. Only with your expert input and review can we get the the most out of generative AI.

Happy designing,
Phil 🙌