[PREP Research Review] The world needs a 'premortem' on the use of generative AI in education [Brookings Institution]

You want to enjoy the world of reading more, but you don't know what to read—for you, I offer the 'PREP Research Review'.

In this project, I aim to introduce the summaries and key points of research that I have read and found educational in an easy-to-understand way, helping you with your own reading, as well as writing for note or blogs.

Specifically, I will use the PREP method (Point, Reason, Example, Point) to convey the appeal of the research in an easy-to-understand manner.

The article I am introducing this time is 'The world needs a ‘premortem’ on generative AI and its use in education,' published by the Brookings Institution.

Actually, at first, I wasn't really that interested in this topic. However, after reading the arguments and examples written in this article, and actually trying to use AI tools in my own classes little by little, I became convinced that 'this might change education significantly'.

Point

A 'premortem' is essential to grasp the light and shadow of AI 'in advance'!

As AI begins to spread widely in educational settings, attention tends to focus on the bright aspects, such as 'streamlining lesson preparation' and 'automatic optimization of learning content'.

However, the Brookings Institution article emphasizes that a 'premortem' (pre-verification), which examines the disadvantages and risks of this new technology first, is what is truly important.

This is because, while new technologies spread rapidly due to their innovation, there is a common pattern where problems that were not visible in the early stages become serious a few years later.

In particular, generative AI can handle a wide range of tasks such as writing and image generation, and has the potential to have an impact that fundamentally changes the way humans learn and work.

While keeping our eyes on the benefits, we must think about 'how we will deal with it if a problem occurs' from now on, regarding the challenges we have not yet fully anticipated.

This can be said to be the main argument of this article.

Reason

To avoid repeating the 'initial enthusiasm' and 'subsequent harm' seen with the spread of smartphones

The Brookings Institution article symbolically tells the story of when the iPhone appeared in 2007.

When smartphones were first released to the world, their innovation as a 'computer in your pocket' attracted a lot of attention in the field of education as well.

Indeed, it is true that constant connection to the Internet became possible, and the hurdles for learning content and research-based learning were significantly lowered. Also, learning methods using multimedia increased, and for learners with disabilities, voice input and screen reading functions were a big step forward.

However, after a few years of that 'enthusiasm,' a serious problem called 'smartphone addiction' erupted.

Students becoming so absorbed in social media and games that they cannot focus in class, or parents and teachers alike falling into a state where they cannot let go of their smartphones...

Furthermore, as negative impacts on mental health have begun to be pointed out, 'how to restrict smartphones' has become a major topic of discussion around the world.

This trend is by no means limited to smartphones; it has been repeated throughout the history of the introduction of past technologies, such as television and social media.

'Expectations are high at first and adoption spreads rapidly, but later, issues that were never anticipated erupt'—this is not someone else's problem for us either.

Because generative AI holds the potential for more diverse applications than smartphones, the problems that could arise behind the scenes may also be more complex than we imagine.

That is precisely why a 'pre-mortem,' which anticipates negative aspects in advance and takes countermeasures, is indispensable.

Example

Two major questions presented by the 'pre-mortem' task force

The Center for Universal Education (CUE) at the Brookings Institution spent about two years launching a task force to anticipate the 'risks that generative AI brings to education.'

There are two main core questions that this task force is addressing.

The first is, 'What are the potential risks that generative AI poses to children's learning and development?'

The second is, 'Knowing those risks, how can we utilize generative AI in educational settings to improve the quality of learning?'

For example, by having AI automate part of a teacher's work, the time spent on lesson preparation and grading assignments could be reduced, potentially allowing teachers to devote more energy to communicating with students.

However, at the same time, if lesson design and feedback are left entirely to AI, there is a risk that the 'intuition for where students stumble' and the 'ability to feel the atmosphere of the classroom firsthand' that teachers should acquire will not be developed.

I also tried creating teaching materials using AI on a trial basis, and while the materials were created smoothly at first glance, I often found myself wondering, 'Is it really okay to give these problems to these children?' or 'Does the explanatory text proposed by the AI match the children's actual level of understanding?'

In short, I feel that the balance between 'room to exercise professional expertise as a teacher' and 'the parts covered by AI' is extremely delicate.

Also, from the student's perspective, they might be able to finish reports quickly using AI, but the question of 'Will this really help me develop my own skills?' remains.

As such, since the merits and demerits are two sides of the same coin, the introduction of AI in educational settings must be approached with caution.

Moreover, according to a UNESCO survey, countries that have established clear guidelines or training systems regarding the use of educational technology, including AI, are still in the minority.

Organizations like Teach AI have presented principles such as 'human judgment should ultimately take precedence when introducing AI,' but globally, we are still in a stage of groping for how it should be utilized.

In reality, I also hear of cases that run to extremes, such as 'banning it because we don't know how to use it' or 'problems erupting after introducing it freely.'

That is precisely why initiatives like this task force, which aim to 'anticipate problems in advance and then consider the best ways to use it,' should be highly instructive.

Point (Reiteration)

Humans are the 'agents of education'—which is exactly why we should be actively creating rules for AI usage now.

In summary, I feel that the 'premortem' approach presented in the Brookings Institution article is a key point for making education in the coming AI era better.

As we can see by looking back at the history of technology, while new tools trigger 'surprise and enthusiasm,' they often bring about unforeseen negative consequences as time passes.

To avoid repeating those 'regrets,' we need to think about preventive measures from this stage with the perspective of 'what will we do if this happens?'

On the other hand, what I personally feel is the positive aspect that 'AI has the potential to enrich education.'

That is why I believe we should actively think about 'how to use it' rather than 'banning AI entirely,' as some suggest.

Drawing a line between the work teachers should do and the work that can be left to AI, while realizing 'ease of learning,' 'ease of teaching,' and 'proactive learning' for both teachers and students—isn't this a very rewarding theme?

For example, if AI can summarize large amounts of information or supplementary materials in a short time, we can imagine new teaching styles where teachers use that as a base to provide customized instruction for each student.

If you read this article and thought, 'I want to know more about the benefits and risks of generative AI' or 'I want to know specifically what to be careful about when introducing it,' please be sure to check the original text and related resources.

Education is inherently something that humans create as the primary agents.
However, if those humans can skillfully draw out the power of AI, it may open up creative learning spaces that we have yet to imagine.

I strongly feel that whether or not such a future becomes a reality depends on how much we ourselves take the initiative to implement measures and get involved in rule-making.

Kyu-san's Bookshelf: For those further interested in research

If you have read this article to the end and are interested in 'The world needs a ‘premortem’ on generative AI and its use in education,' please be sure to see the original text via the link below.

You are sure to find new discoveries.