AI is The Technology Schools Never Asked For

How "Arrival Technologies" Reveal the Limits of Educational Planning

Justin Reich is an edtech researcher at MIT. His book, Failure to Disrupt, landed just as the pandemic was in full swing. He argued persuasively that educational technologies rarely achieve massive change and, if they do have a sizeable impact, it usually happens more slowly than the “tech of the month” hype cycle predicts. Given his earned skeptical outlook about transformative educational technologies, I was eager to follow his thoughts on the introduction of generative AI and its effect on schools. In a recent paper, Reich and co-author Jesse Dukes help explain why AI feels so different from other educational technologies and why some schools may end up thriving while others are drowning.

Reich, Justin, and Jesse Dukes. Toward a New Theory of Arrival Technologies: The Case of ChatGPT and the Future of Education Technology after Adoption. MIT, 2024.

What Just Happened?

A recurring theme you will hear when talking about AI and schools is how it came out of nowhere. Though OpenAI was founded in late 2015 and ChatGPT models were available to play with online as early as 2020, most would agree that November 2022 was the 1492 moment for generative AI.

Essentially dropped from the heavens, the public release of ChatGPT was a "where were you when" moment for educators. I still remember watching text unfurl at warp speed in response to some mindless prompt about writing a poem in the style of Shakespeare. Almost instantly, schools began to see AI text infiltrate student academic assignments, and the disruption hasn't stopped since.

If we went back in time, teachers would uniformly veto giving free and unlimited access to generative AI technologies to students. There was no training. No discussion. No real understanding of what these models did or were capable of doing. No edtech consultant would authorize rolling out such a powerful and buggy product in schools without committees, piloting, evaluation, and more … It would be insane.

And yet that is exactly what happened.

We didn't ask for this said every teacher everywhere. We were completely unprepared.

What now?

The Framework That Explains Our Current Predicament

Reich and Dukes offer a simple but powerful framework by explaining the difference between adoption technologies and arrival technologies.

Adoption technologies are the ones schools choose to implement after careful consideration - think smartboards, learning management systems, or Chromebooks. Under these circumstances, schools form committees, pilot programs, evaluate options, train staff, and implement gradually.

Arrival technologies bypass all of that. As Reich and Dukes put it:

ChatGPT arrived. It entered schools without evaluation, assessment of risks and benefits, training for educators, or any other adoption steps that historically would have been considered indispensable to effective technology integration.

The distinction is critical. Arrival technologies enter schools through "the unguided, spontaneous actions of students, teachers, and technology vendors," creating immediate challenges for everyone involved. In this case, the technology vendors are huge Silicon Valley corporations who have simply assumed that student use of AI is beneficial without any pedagogical expertise. Basically, schools were left to fend for themselves.

The adoption/arrival framework seems obvious in hindsight, but it helps reshape the conversation around school response to AI with more empathy. Even under ideal circumstances, introducing generative AI into schools would require extensive planning, training, and advanced preparation. Arriving without warning inevitably created chaos that no school system was designed to handle.

Why AI Really Is Different

As the authors point out, previous technologies - calculators, mobile phones, Wikipedia, social media - had some arrival characteristics, but they trickled into schools slowly over years, giving educators time to adapt. ChatGPT and generative AI LLMs are fundamentally different for several reasons:

Infrastructure was already in place

The pandemic left nearly every secondary school with ubiquitous student digital access. When ChatGPT launched, almost every student already had the infrastructure to use it immediately, whether through a school issued or personal computer.

Adoption was instant and massive

ChatGPT became the fastest-adopted web platform in history - reaching 100 million users in just 60 days compared to Instagram's 2.5 years or TikTok's 9 months. Almost instantly, students made up a significant percentage of weekly ChatGPT users. That trend shows no signs of slowing down.

I would add a third difference.

The capability gap is unprecedented

AI's raw abilities dwarf anything that came before, and, as of today, dozens of other generative AI tools have flooded the market, all with freely available versions for students to choose from. Unlike the other technologies mentioned, AI applies to every subject and the speed with which new models, features, and modalities have improved is unlike anything we have ever experienced.

These three factors combined to create something I haven’t seen in my thirty years in school: a disruptive technology that arrived simultaneously in every classroom, with capabilities that apply to virtually every assignment, and infrastructure already in place for immediate use.

The Tidal Wave Continues

The MIT researchers emphasize that generative AI is an arrival technology that will continue to "arrive" whether we like it or not - it's essentially a technology we can continue to prepare for but not one we really have a choice whether or not to "adopt" in the traditional sense. We are currently at the mercy of Big Tech.

AI is embedding everywhere

Beyond student and teacher experimentation, AI will arrive embedded in tools schools already use. Google Docs, gradebooks, learning management systems - they're all adding AI features whether schools asked for them or not. Districts that spent years choosing between Google and Microsoft will suddenly find their carefully selected tools transformed overnight.¹

This is already happening at lightening speed. It's almost pointless to try to keep up with the latest AI tool, model or integration. Google just added AI Search mode, Claude recently added their Deep Research component, and sound, image, and video platforms crop up daily. I just learned this morning that o3 Pro, a new Open AI model, was released yesterday and, in the words of a frequent AI writer, “it’s easily the best model in the world.”

ChatGPT may be the shorthand reference point for generative AI, but LLMs are baked into almost every digital platform we will be using going forward. Regardless of whether we choose to formally “adopt” AI technologies in schools or not, our students will have access to them. That is not going to change.

The Education Divide

Here's where the arrival technology framework reveals its most troubling implications.

Reich and Dukes note that well-resourced schools have always been better positioned to benefit from new technologies - they have the "financial, social, and technical capital to support this developmental improvement process." But arrival technologies make this problem worse.

Wealthier schools can adapt on the fly

As schools are already figuring out, mitigating AI's harms while capturing its benefits is "a poorly understood, hugely time-consuming task." Wealthy schools can manage "procurement-on-the-fly" - writing policies, running pilots, testing multiple services. Under-resourced schools? They're drowning. These schools are forced to react without sufficient resources.

As one urban teacher in southern California told the researchers: "AI hasn't provided ANY benefits to me as an English teacher. And it hasn't forced me to change much. It's just another thing I have to deal with. One of too many. I'd love to take a professional learning day on ChatGPT, but our district has zero subs."

This simply underscores how much the AI revolution is likely to impact schools differently around the country and around the world. It will be an important point to remember when evaluating the success or failure of AI pilot programs and initiatives. What works in some places might fail spectacularly in others, making it far more difficult to draw definitive research conclusions about AI's potential to either aid or undermine learning goals.

Another Arrival Technology Around the Corner?

Perhaps most exhausting for educators: "ChatGPT is best understood as the first widely - and instantly - embedded AI-powered arrival technology, it is unlikely that it will be the last."

Given that teachers and schools are still reeling from the fallout of ChatGPT and its successors over the past two and a half years, the thought that another AI technology might appear just around the corner is exhausting.

But this is the reality of arrival technologies. They don't ask permission, follow procurement schedules, or wait for school readiness.

Reich and Dukes conclude their paper with pragmatic wisdom and a potential call to action:

"If new tools enhance learning and wellbeing, let's use them to the best of our ability, guided by robust theory on maximizing the benefits of arrival technologies. And if innovations entering our schools harm our students, let's fight."

Amen.

1. Though not referenced in the paper, Marc Watkins at Rheotrica wrote precisely about this issue in a recent post.