London, November 2023
A year ago, the most disruptive force in education arrived without warning, a syllabus, or a single staff meeting to prepare for it. ChatGPT reached an estimated 100 million users within two months of its launch, the fastest adoption of any consumer technology on record. It walked straight into the classroom — and the classroom, it turns out, had no idea what to do.
Twelve months on, the response has lurched from prohibition to embrace, exposed a near-total vacuum in official guidance, and surfaced a question far bigger than cheating: when a machine can write a passable essay in seconds, what exactly is a student being taught to do? According to Ewance — a challenge-based learning platform where students solve real-world challenges tied to their field of study — 2023 is the year that question stops being hypothetical.
How did schools go from banning ChatGPT to embracing it in five months?
The first instinct was to slam the door. In January 2023, New York City’s Department of Education — the largest school district in the United States — blocked ChatGPT on its devices and networks, citing "negative impacts on student learning, and concerns regarding the safety and accuracy of content." Other districts and universities followed with bans of their own.
The door did not stay shut. On 18 May 2023 — barely five months later — the same NYC department reversed course, with schools Chancellor David Banks writing that the system had moved with "initial caution" but would now work to integrate the technology and support educators and students exploring it. The fastest U-turn in recent education policy told its own story: you cannot ban your way out of a tool this useful and this available.
The ban-and-reverse whiplash wasn't a failure of nerve — it was a collision with reality. You can block a website on the school network. You cannot block it on a teenager's phone. The only durable answer is to change what and how we teach.
— Ewance's founder
The technology raced ahead of the rulebook
The reversals happened in a governance vacuum. In September 2023, UNESCO published the first-ever global guidance on generative AI in education — and in doing so revealed how unprepared institutions were: fewer than 10% of schools and universities worldwide had any formal guidance on the use of AI. UNESCO urged governments to regulate its use and suggested a minimum age of 13 for classroom use.
That is the heart of the problem. A technology adopted by tens of millions of students and teachers was operating, in nine schools out of ten, with no rules at all.
Teachers, not students, led the adoption
The popular image of 2023 was of students secretly outsourcing their homework. The data tells a more surprising story: the early adopters were the teachers. Surveys for the Walton Family Foundation by Impact Research found that within two months of ChatGPT’s launch, 51% of teachers reported using it, with 40% using it at least weekly — against just 22% of students using it weekly or more.
Adoption deepened through the year. By the summer, 63% of teachers said they had used ChatGPT for their job — up 13 points from 50% in February — and 84% of teacher users said it had a positive impact on their classes. Parents were broadly onside: 64% said schools should allow students to use ChatGPT for schoolwork, including 28% who said schools should actively encourage it.
The lazy narrative was "students will cheat." The reality was teachers reaching for it first — to plan lessons, draft materials, and differentiate for their classes. When the people grading the work adopt the tool faster than the people doing the work, the old assessment model is already obsolete.
— Ewance's founder
Are students cheating, or figuring it out for themselves?
Students, for their part, were adopting with more caution than the headlines implied. A Pew Research Center survey conducted in the autumn and published in November 2023 found that two-thirds of US teens had heard of ChatGPT — but only about 1 in 5 of those who had heard of it (19%) had actually used it for schoolwork. Among the most aware — teens who had heard "a lot" about the tool — that figure rose to 36%.
The picture, in other words, was neither a cheating epidemic nor a non-event. It was a generation quietly working out, without much guidance, where the line should sit.
The detection arms race nobody wins
Faced with uncertainty, many institutions reached for a technological fix. In April 2023, Turnitin switched on AI-writing detection across its products, claiming it could identify AI-generated writing with up to 98% confidence. But detection quickly proved a fragile foundation: false positives risked penalising innocent students, accuracy was contested, and the underlying generators were improving faster than the detectors chasing them.
The deeper issue is that detection treats the symptom. If the only thing standing between a student and a finished assignment is the will to paste a prompt, the problem was never the chatbot — it was an assessment that asked for something a machine can now produce.
You can't detect your way back to 2019. Every hour spent policing whether a paragraph was written by a person is an hour not spent asking a better question: what can this student actually do that an AI can't?
— Ewance's founder
The real question isn't cheating — it's relevance
That is the shift 2023 forced into the open. If a chatbot can produce a competent five-paragraph essay on demand, the value of teaching — and assessing — the five-paragraph essay collapses. The premium moves to the things AI cannot fake: applying knowledge to a messy, real problem; exercising judgment; producing work whose quality is visible in the doing, not just the document.
It is a case for learning that is harder to outsource because it is rooted in genuine application — students solving real challenges tied to their field, where the process and the result are the proof. The technology that broke the take-home essay may, paradoxically, push education toward something more authentic than the essay ever was.
What comes after the chatbot?
The most useful response to 2023, in Ewance's view, isn't a better detector — it's better assignments. The platform was built on the premise the year exposed: when a machine can produce the document, the value moves to the doing. Launched in 2023, Ewance runs challenge-based projects with university students — real problems tied to their field, where the work itself is the proof, not a paragraph a chatbot could have written.
Its perspective is grounded in the field, not the feed: Ewance's founder has spent more than a decade working at the intersection of education and industry, and the platform's read on 2023 was formed alongside the students, educators, and recruiters living the change.
You can't police your way back to the old essay. You design forward — toward work that's worth doing because a student, not a model, has to do it. We've spent more than a decade in that room, and 2023 only sharpened the lesson.
— Ewance's founder
Recommendations
Ewance's founder offered the following for schools, students, and parents heading into 2024:
- Stop policing, start redesigning. Assessment that can be completed by a chatbot needs rethinking, not better surveillance.
- Teach with AI, not against it. The teachers already using it are modelling the literacy students will need at work.
- Value application over recall. The durable skills are the ones demonstrated on real problems, where AI is a tool rather than the author.
- Close the guidance gap. With fewer than one in ten institutions offering formal rules, clear, honest policy is the single fastest win available.
The chatbot didn't break education. It exposed how much of education was already testing the wrong thing. The schools that treat 2023 as a wake-up call, rather than a threat to be blocked, are the ones whose students will be ready for what comes next.
— Ewance's founder
Sources
This release synthesises publicly available research and reporting published on or before November 2023; NYC, Walton/Impact, Pew and Turnitin data is US-based, UNESCO and the adoption figure global. Ewance conducted no proprietary survey for this release.
- UBS analyst note (February 2023) — ChatGPT reached ~100M monthly active users within roughly two months of launch, the fastest-growing consumer application on record.
- New York City Department of Education — blocked ChatGPT on school devices and networks (January 2023); ban rescinded 18 May 2023 (Chancellor David Banks).
- UNESCO — "Guidance for generative AI in education and research" (September 2023): first global guidance of its kind; fewer than 10% of schools and universities have formal guidance on AI; suggested minimum age of 13 for classroom use.
- Walton Family Foundation / Impact Research — (February 2023) 51% of teachers had used ChatGPT (40% at least weekly) vs 22% of students weekly or more; (June–July 2023) 63% of teachers had used it for their job (up from 50%), 84% of teacher users reported a positive impact, 64% of parents said schools should allow its use (28% encourage it).
- Pew Research Center (published 16 November 2023; 1,453 US teens, 26 Sept–23 Oct 2023): two-thirds of US teens had heard of ChatGPT; ~1 in 5 (19%) of those aware had used it for schoolwork; 36% among teens who had heard "a lot."
- Turnitin (April 2023) — launched AI-writing detection across its products, claiming up to 98% confidence.

