Why Banning AI in Classrooms Misses the Point
Key insights
- Banning AI from classrooms repeats the same mistake as resisting calculators and GPS, according to this IBM engineer
- AI tutors could level the playing field by giving every student access to personalized, patient instruction through a browser
- The critical skill for an AI-powered future is not prompt engineering but critical thinking, because AI will produce wrong answers that humans must catch
This is an AI-generated summary. The source video includes demos, visuals and context not covered here. Watch the video → · How our articles are made →
Read this article in Norwegian
In Brief
Jeff Crume, IBM Distinguished Engineer and adjunct professor, argues that banning AI from classrooms is like insisting students write in cursive in a world that runs on keyboards. Generative AI, technology that can produce text, images, and code from simple instructions, is already reshaping how students work. Rather than treating it as a threat, Crume contends that educators should embrace it to foster critical thinking, personalized learning, and equity. The video traces how education has always adapted to new tools, and asks why AI should be any different.
For related coverage, see How Alpha School Uses AI Tutors to Teach Kids 10x Faster, From Estonia to India: Schools Bet Big on AI, and Anthropic's CEO on Why Coding Skills Still Matter.
The central claim
Crume opens with a candid admission: the first time he saw what a chatbot could do, he knew assigning traditional essays was pointless (0:00). He would only be grading the AI's output, not the student's thinking. Most educators responded by banning AI entirely (0:16), but Crume argues this approach is doomed. Detection tools will always fall behind as chatbots keep improving (0:27).
His core metaphor: "The train has already left the station. We can either get on board or get run over by it" (0:44).
The historical pattern
Crume illustrates his point with skills that once seemed essential but faded as technology advanced (1:16):
- Cursive writing and penmanship were formal courses a generation ago. Today most writing happens on keyboards.
- Memorization of facts like the periodic table consumed significant instructional time. Crume recalls memorizing all 106 elements (2:07). Now any detail can be looked up in seconds.
- Complex arithmetic by hand was drilled repeatedly, but calculators have been standard in college courses for decades.
- Map reading remains useful, but GPS navigation is how most people get around today.
The argument is not that these skills are worthless, but that instructional time is limited. Educators must choose which abilities best prepare students for the world they will actually enter (4:03).
What AI brings to the classroom
Crume lists several concrete advantages of integrating AI into education:
Just-in-time learning. Students can get a focused explanation of any topic at the moment they need it, rather than waiting for the next scheduled lesson (6:56).
Personalized tutoring. An AI tutor, a system that provides one-on-one educational support adapted to each learner, is "infinitely patient" (7:45). It can try different explanations when a student is stuck, even when the best teacher might run out of patience or get pulled away by other students. This connects to what schools like Alpha School are already experimenting with.
Editing and feedback. Beyond spell-check, AI can explain why a grammar rule applies, turning corrections into teaching moments (8:11).
Accessibility. Students with disabilities can benefit from text-to-speech, speech-to-text, and other AI-powered tools that make learning materials more accessible (8:49).
Debate preparation. Students can use AI as a research assistant, then defend their positions in live debates. The AI can also poke holes in their arguments, sharpening their reasoning (10:22).
Educational equity. A cloud-based AI tutor only requires a browser and an internet connection, potentially giving students in under-resourced areas access to high-quality instruction (10:40).
Opposing perspectives
The cheating problem is real
Crume acknowledges that many educators banned AI because students would "let AI do all the work" (0:22). Critics argue that foundational skills like essay writing develop more than just the final product. They train the process of organizing thoughts, building arguments, and revising drafts. If students skip that process entirely, they may miss learning outcomes that no amount of AI integration can replace.
Access is not equal
The equity argument assumes reliable internet access and devices for all students. In many parts of the world, and in many under-funded school districts, that assumption does not hold. Shifting education toward AI tools could widen the gap for students who lack basic digital infrastructure. Countries like Estonia and India are investing heavily in AI education, but their approaches show how uneven adoption can be.
The calculator analogy has limits
Crume's historical parallels are intuitive, but calculators do not hallucinate, the term for when AI generates plausible-sounding but factually incorrect information. A student who trusts a calculator's output is safe. A student who trusts an AI chatbot's output without verification may learn things that are simply wrong. Critical thinking is also harder to teach than arithmetic, making the transition more complex than the analogy suggests.
How to interpret these claims
Crume presents a persuasive narrative, but several factors deserve consideration.
Stories, not studies
The argument relies on historical analogies (cursive, calculators, maps) rather than controlled studies of AI in classrooms. These parallels are intuitive, but they do not prove that AI integration produces better learning outcomes. The shift from map reading to GPS did not change how people develop spatial reasoning. But the shift from essay writing to AI-assisted work may change how students learn to think.
No measurable evidence cited
Crume does not reference specific studies, pilot programs, or measurable outcomes. The video presents a philosophical position rather than hard data. Stronger claims would include results from schools that have tried AI-integrated curricula and compared them against traditional approaches. Without this, the argument rests entirely on analogy and intuition.
The industry perspective
As an IBM Distinguished Engineer presenting on IBM's official YouTube channel, Crume has a professional interest in promoting AI adoption. This does not invalidate his arguments, but the perspective comes from inside the AI industry, not from education research. Independent studies from education institutions would carry more weight.
What stronger evidence would look like
Controlled comparisons of AI-integrated versus traditional classrooms measuring learning outcomes over multiple semesters. Student performance data from schools that adopted AI tutoring tools. Longitudinal research showing whether AI-trained students retain critical thinking skills in AI-free environments.
Practical shifts
Crume suggests rethinking what to teach, not simply adding AI tools (13:02):
- Essays to debates. Replacing some writing assignments with structured debates tests critical thinking, communication, and quick reasoning in real time.
- Memorization to principles. Understanding why elements are grouped on the periodic table builds reasoning that transfers to new problems (13:53).
- Arithmetic to algebra. Time saved on long division can go toward higher-order mathematics. Analysis and logical thinking matter more than repetition.
Crume also emphasizes two prerequisites: AI literacy, understanding what AI can and cannot do (12:10), and ethics, grasping that "just because you can do something doesn't mean you should" (12:40).
His closing argument is blunt: "What boss is going to say do this, but don't use AI?" (15:07). Training students without AI, he argues, is training them to live in the past (15:31).
Glossary
| Term | Definition |
|---|---|
| Generative AI | AI systems that can create new text, images, code, or other content from instructions. ChatGPT is the most well-known example. |
| Chatbot | A software application that simulates conversation with users, typically powered by a language model. |
| AI literacy | The ability to understand what AI can do, what it cannot do, and how to use it responsibly. |
| Just-in-time learning | Getting focused instruction on a specific topic at the exact moment it is needed, rather than following a fixed curriculum schedule. |
| Personalized learning | Education that adapts pace, style, and content to each individual student's needs and abilities. |
| AI tutor | An AI system designed to guide students through learning material, answer questions, and adapt explanations to the learner's level. |
| Hallucination | When AI generates plausible-sounding but factually incorrect information. A key reason critical thinking matters when using AI tools. |
| Responsible AI | The practice of developing and using AI in ways that are ethical, transparent, and mindful of potential harms. |
| Accessibility | Making tools and content usable by people with different abilities, including those with visual, hearing, or cognitive disabilities. |
| Deskilling | The gradual loss of skills when tools automate tasks that humans used to perform manually. A concern raised by AI critics. |
Sources and resources
Want to go deeper? Watch the full video on YouTube →