Beyond pattern matching - Nurturing the human intelligence AI cannot replace

Oct 23, 2026

Upload by Mia

A keynote presentation exploring how AI intersects with education and human intelligence, focusing on the question of whether learners are prepared to thrive with AI or pushed into dependency. It reviews evidence from student surveys on assessment understanding, adoption patterns, and risks of unmanaged “shadow AI,” while noting gaps in teacher training and organizational policy. It also highlights unequal AI uptake across subjects and the cognitive impacts found in research.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages48

Reading Time2 h 1 min

AccessFree to download

AI and Education: where are we?

What is Human Intelligence?

Enhancing Human Intelligence through AI

The visible gap: data on AI and learning

The real risk is not AI; it is unmanaged AI

Equity: who benefits?

The cognitive concern, in evidence

Closing Thoughts

Beyond pattern matching: Nurturing the human intelligence AI cannot replace Prof. Rose Luckin CBE Professor Emerita UCL CEO & Founder EVR Ltd For all your AI needs: we help you realise the opportunity of AI Today, we will explore: 1. The profound question we face: AI and Education: where are we? 2. What is Human Intelligence? 3. How do we enhance Human Intelligence through AI 4. Closing Thoughts. “Learn Fast Act More Slowly” Key Question Are we preparing learners to thrive with AI, or are we inadvertently creating dependency on AI? Cambridge Schools Conference, UK · 15 to 16 July 2026 · Licensed under CC BY-NC-ND 4.0 What is the data telling us about AI and Learning? Cambridge Schools Conference, UK · 15 to 16 July 2026 · Licensed under CC BY-NC-ND 4.0 The visible gap A national UK student survey, March 2026 Generative AI did not create the gap between submitting work and understanding it. It made that gap impossible to ignore. 47% worry their grades do not reflect what they actually know. Wonkhe (2026), Trained to Stop Learning. Survey of 1,055 UK students across 52 providers. 38% admit submitting work they could not fully explain. Cambridge Schools Conference, UK · 15 to 16 July 2026 · Licensed under CC BY-NC-ND 4.0 The real risk is not AI; it is unmanaged AI Student adoption More than nine in ten UK students now use AI tools, up from two-thirds just one year earlier. Teacher preparedness 67% of US teachers received no AI training last year (Gallup, 2025) . Only 25% were taught what AI actually is. Policy and governance Only 44% of executives report that their organisation has an AI policy, up from 10% in 2023. More than one in four companies have no policy and no plans for one (Littler, 2025) . Shadow AI The majority of AI tools entering schools and workplaces are brought in by staff or students without employer knowledge or approval (Microsoft Work Trend Index) . The global policy response has been fragmented Technology companies operate globally. Policy does not. Users are not waiting for researchers to tell them something is wrong Ganguli et al. (2026), Anthropic Claude User Survey: Hopes and Concerns About AI. 80,508 respondents, 159 countries, 70 languages. Cambridge Schools Conference, UK · 15 to 16 July 2026 · Licensed under CC BY-NC-ND 4.0 E q u i t y , w h o b e n e f i t s ? AI adoption is not evenly distributed across subjects. Aggregate adoption: 60.6 per cent. Subject-level adoption tells a different story. Above the 6 0 . 6 % average Mathematical Sciences 77.8% Computing 75.3% Physical Sciences 72.9% Engineering and Tech 70.0% Business and Management 67.7% Below the 6 0 . 6 % average Creative Arts and Design 46.4% Humanities and Liberal Arts 41.6% Language and Area Studies 36.8% Add gender, ethnicity and socio-economic status, and the gaps compound. Morgan et al., PAQ Wave 1, n=5,500 incoming undergraduates, September 2025. Cambridge Schools Conference, UK · 15 to 16 July 2026 · Licensed under CC BY-NC-ND 4.0 The cognitive concern, in evidence We are right to ask the question. The early evidence does point to a real risk, but also to a real path through it. Bastani et al. (2025), PNAS Students who practised with a general-purpose AI chatbot scored 17 per cent lower on closedbook assessments than peers without AI access. Stadler, Bannert and Sailer (2024), Computers in Human Behavior Students using a chatbot produced lower-quality reasoning and argumentation, with measurably reduced mental effort, compared to peers using a search engine. Kosmyna et al. (2025), MIT preprint Reduced brain activity and weaker recall in students using AI for essay writing, with the effect persisting after the tool was removed. Cambridge Schools Conference, UK · 15 to 16 July 2026 · Licensed under CC BY-NC-ND 4.0 What is AI and how does it impact on education?

FAQs

Why does the presentation claim the real risk is unmanaged AI rather than AI itself?

It argues risk stems from how AI enters schools and workplaces through staff or students without approval, creating “shadow AI.” It also notes limited teacher training and incomplete policy coverage among executives.

How does the document describe equity in AI adoption across subjects and groups?

AI adoption is presented as uneven, with mathematical and computing subjects showing higher adoption than creative arts, humanities, and language/area studies. The document further states that adding gender, ethnicity, and socio-economic status can compound these gaps.

Oct 23, 2026

Upload by Mia