Nick Clegg在2025年1月离开Meta后，于2026年3月加入两家AI公司董事会：资料中心公司Nscale与教育新创Efekta。Efekta源自EF Education First，提供可依学生能力调整并向教师回报进度的AI助教；该平台目前约服务400万名学生，主要分布于拉丁美洲与东南亚。Clegg认为，教育将是最早被AI显著改善的场域之一，因为AI可把传统课堂难以实现的一对一教学，扩展为大规模、互动式、适性化教学。
对于AI本身，Clegg拒绝「末日论」与「救世论」两端夸大叙事，主张这项技术同时「非常多才」也「非常愚笨」：在写程式等任务上格外强大，在许多其他任务上则明显无效。他认为，学生本来就会使用聊天机器人，因此重点不在禁止，而在制度化纳入教学。风险方面，他特别指出儿童与脆弱成年人，可能对具有人形化、全天24小时陪伴特征的agentic AI产生情感依附，因此支持更严格的年龄分级与平台层级管制。
他对「superintelligence」一词尤其怀疑，认为即使在同一机构内问3个人，也可能得到3种定义；这种模糊愿景常被用来吸引顶尖研究者。更核心的问题是权力集中：AI一方面提升个人能力，另一方面又因大型语言模型基础设施极度昂贵，而把控制力推向更少数公司。他预测现行路径终将出现洗牌，因为产业不可能长期每年持续投入1,300亿英镑。基于相同理由，他批评欧盟AI监管框架削弱欧洲创业者，并主张开源比高度封闭的专有模式更能民主化技术。
After leaving Meta in January 2025, Nick Clegg joined the boards of two AI companies in March 2026: data-center firm Nscale and education startup Efekta. Spun out of EF Education First, Efekta sells an AI teaching assistant that adapts to student ability and reports progress to teachers; the platform now serves about 4 million students, mainly in Latin America and Southeast Asia. Clegg argues that education will be one of the earliest sectors radically improved by AI because it can scale the one-to-one attention that traditional classrooms usually cannot provide.
On AI itself, Clegg rejects both doomerism and boosterism, arguing that the technology is simultaneously “very versatile” and “very stupid”: exceptionally strong in tasks like coding, yet plainly weak in many others. He says students will use chatbots anyway, so the issue is not prohibition but structured integration into teaching. On risk, he highlights children and vulnerable adults who may become emotionally attached to agentic AI with avatars or constant 24-hour availability, and he therefore favors stricter age-gating and platform-level controls.
He is especially skeptical of “superintelligence,” noting that asking 3 people in the same organization can produce 3 different definitions; in his view, the vagueness often helps firms recruit elite researchers. His larger concern is a power paradox: AI can empower individuals while concentrating control in fewer companies because large-language-model infrastructure is extraordinarily expensive. He expects a market shakeout, arguing that the sector cannot keep spending £130 billion a year indefinitely. For the same reason, he criticizes EU AI regulation as self-harming for European founders and argues that open source is a better route to technological democratization than tightly closed proprietary systems.