Google DeepMind Chief Operating Officer Lila Ibrahim posed a pivotal question in a recent CNBC appearance: Will AI make or break education? Her insights, drawn from ongoing pilots and research, suggest the technology holds transformative power if wielded responsibly. Speaking on “The Tech Download,” Ibrahim highlighted opportunities alongside risks, urging immediate education on ethical AI use (CNBC).

Ibrahim’s perspective stems from her extensive background in education technology. As former president and COO of Coursera, she scaled online learning to millions worldwide. Now at DeepMind since 2018, she oversees operations, ethics, and partnerships, chairing initiatives on AI responsibility. Her personal drive traces to her sister with cerebral palsy, who earned two master’s degrees despite barriers, and her neurodiverse twin daughters, fueling advocacy for personalized AI tutoring (CNBC).

DeepMind’s efforts emphasize pedagogy over rote answers. “To realize this potential, AI learning tools must help learners cultivate deep understanding, not just deliver quick answers. They must ignite curiosity and engage learners in a process of discovery—not offer a shortcut,” Ibrahim wrote in a Google blog post (Google Blog).

Northern Ireland Pilot Reveals Real-World Gains

A six-month trial with Northern Ireland’s Education Authority C2k program equipped 100 teachers with Gemini and Google Workspace tools. Participants reported saving up to 10 hours weekly on administrative tasks, redirecting time to student interaction. Over 600 unique use cases emerged, from drafting parent letters to generating tailored lessons. Chris Lowe, Head of ICT at Ashfield Boys’ High School, said, “The time I saved using Gemini fundamentally allows me to do the job I want to do—and that is to teach” (Google DeepMind Blog).

Alistair Hamill, Head of Geography at Lurgan College, leveraged NotebookLM’s MindMap for a neurodivergent student, visualizing complex material. An ICT coordinator at Rowandale Primary called Gemini “game-changing” for personalized lessons and creative prompts. Damian Harvey, Interim Head of C2k, stressed partnership: “I believe educators need to embrace the opportunity” (Google DeepMind Blog).

Ibrahim, visiting the schools, praised teachers’ ingenuity: “I was struck by the teachers’ ingenuity and the efficiencies they were able to achieve” (Google DeepMind Blog).

Tools Tailored for Learning Science

DeepMind’s LearnLM family powers education-focused models integrated into Gemini, NotebookLM, and Google Classroom. These enable guided conversations, quizzes, and immersive experiences grounded in pedagogy. In a Les Echos interview, Ibrahim noted continuous improvement: “More the product is used, more it improves, it’s a cycle of continuous improvement” (Les Echos).

Ben Gomes, Google’s education lead, added that a “large part of the use of information is learning.” Tools like Experience AI, partnered with Raspberry Pi Foundation, have reached educators in 150 countries, equipping 11- to 14-year-olds with AI literacy. Google.org’s $10 million expands it to millions in Europe, Middle East, and Africa (Google DeepMind).

UK government collaboration scales these efforts. DeepMind supports randomized trials showing AI tutoring boosts problem-solving by 5.5 percentage points over human tutors alone. Plans tailor Gemini to England’s curriculum via LearnLM, easing workloads (Google DeepMind Blog).

Scholarships Bridge Access Gaps

DeepMind funds AI Masters scholarships via Martingale Foundation in the UK for 2026, alongside global postgraduate and postdoctoral fellowships at top universities like Oxford and Cambridge. The Pan-African AI for Science Masters with AIMS targets 160 students by 2027. Undergraduate Research Ready placements widen participation (Google DeepMind).

Team4Tech, co-founded by Ibrahim, delivers tech curricula to underserved communities. These initiatives counter equity risks, ensuring AI benefits diverse groups. Ibrahim stresses: “We’re never going to have as much time as we have now” to upskill (CNBC).

Risks loom large: academic integrity, cheating, bias, and over-reliance. Google addresses these through AI literacy, SynthID watermarks for generated content, and assessments like debates that AI can’t replicate. Partnerships rethink evaluation for critical thinking (Google Blog).

Governing the Transformation

Ibrahim co-chairs DeepMind’s Responsibility and Safety Council, evaluating projects against AI Principles. She balances opportunities and risks daily: “Part of my job is to worry every day” (CNBC). In five to ten years, AI ubiquity in education demands adaptation, from Global South pilots in Ghana to advanced markets.

At BETT 2026, Google unveiled Gemini updates for Classroom, including AI Writing Coach with Khan Academy and content verification, prioritizing thinking over answers. Irina Jurenka, DeepMind’s AI education research lead, explores Gemini’s pedagogical adaptations (The 74).

Ibrahim envisions AI maximizing human potential, as with her sister’s achievements. Responsible deployment, she argues, positions education for unprecedented personalization and efficiency.