英国政府目前预计，到2030年，人工智能将为经济增加550亿至1400亿英镑。其核心估算值为1200亿英镑。其《人工智能机遇行动计划》曾称收益将达4000亿英镑，而人工智能大臣卡尼什卡·纳拉扬现已确认，该数字背后的报告是由谷歌委托编制的。
The UK government now expects AI to add £55bn to £140bn to the economy by 2030. Its central estimate is £120bn. Its AI Opportunities Action Plan put the gain at £400bn, and AI minister Kanishka Narayan has now confirmed that Google commissioned the report behind that number.
纳拉扬于10月26日在一份书面议会答复中给出了这两组数字。他于7月成为英国首位内阁级人工智能大臣。保守党影子人工智能大臣本·斯宾塞博士曾询问该计划中4000亿英镑主张的证据来源。
Narayan gave both figures in a written parliamentary answer on 26 October. He became the UK’s first cabinet-level AI minister in July. Dr Ben Spencer, the Conservative shadow AI minister, had asked what evidence sat behind the plan’s £400bn claim.
纳拉扬写道：“4000亿英镑这一数字源自一份公开的2023年报告，该报告由谷歌委托、Public First编制。”
“The £400 billion figure was drawn from a publicly available 2023 report commissioned by Google and produced by Public First,” Narayan wrote.
他表示，政府此后进行了自己的评估。评估发现，人工智能采用和数据投资可使英国生产率增长增加0.4至1.3个百分点。
He said the government had since made its own assessment. It found that AI adoption and data investment could add 0.4 to 1.3 percentage points to UK productivity growth.
他写道：“这些估算存在巨大不确定性，但这相当于到2030年为英国GVA增加550亿至1400亿英镑。”
“These estimates are hugely uncertain, but this would be the equivalent of adding £55-140 billion to UK GVA in 2030,” he wrote.
GVA即毛增加值，衡量一个经济体生产出的价值。该答复并未撤回4000亿英镑这一数字。
GVA, or gross value added, measures the value of what an economy produces. The answer did not withdraw the £400bn figure.
4000亿英镑的来源 谷歌于2023年7月发布了这一较高数字。它来自咨询公司Public First编制的一份经济影响报告。该报告称，“在适当条件下”，到2030年，人工智能可为英国创造超过4000亿英镑的价值。
Where the £400bn came from Google published the higher number in July 2023. It came from an economic impact report that the consultancy Public First compiled. The report said that “under the right conditions”, AI could create over £400bn in value for the UK by 2030.
科技投资人马特·克利福德撰写了该行动计划，政府于2025年1月发布。
Tech investor Matt Clifford wrote the action plan, which the government published in January 2025.
政府给出的较低区间源自其自身模型。科学、创新和技术部在其“人工智能技能提升”计划的解释文件中阐述了该模型。该模型基于经合组织（OECD）的生产率估算，设定了缓慢、中等和快速采用三种情景。
The government’s lower range comes from its own model. The Department for Science, Innovation and Technology sets it out in an explainer for its AI Skills Boost. The model runs three scenarios, for slow, medium and rapid adoption, using OECD productivity estimates.
本·加特赛德为《连线》报道，这是有记录以来政府首次将这两个数字并列呈现。这也是政府首次承认该报告由谷歌委托。《连线》发现，在部长们已掌握较低数字数月后，技能英格兰仍发布了4000亿英镑的数据。采用缓慢，呼吁审查现名为PF Global的Public First告诉《连线》，其2023年模型假设的采用速度比政府现在预期的要快。该公司表示，其目前对2030年的核心估算与政府的数字更为接近，此后增长会更多。该公司还称，在该研究中它做出了自己的判断。
Ben Gartside reported for WIRED that this is the first time on record the government has set the two figures side by side. It is also the first time it has said Google commissioned the report. WIRED found that Skills England still published the £400bn figure months after ministers had the lower one. Slow adoption, and calls for scrutiny Public First, now called PF Global, told WIRED that its 2023 model assumed faster adoption than the government now expects. Its current central estimate for 2030 is much closer to the government’s figure, it said, with more growth after that. The firm also said it made its own judgements in that research.
保守党贵族、里士满的霍姆斯勋爵告诉《连线》，他在2025年1月曾询问部长们该数字的来源。他说他没有得到回复。
Conservative peer Lord Holmes of Richmond told WIRED he asked ministers where the figure came from in January 2025. He got no reply, he said.
担任下院科学技术委员会主席的奇·翁武拉女爵士告诉《连线》，夸大技术所能带来的成果，有削弱变革理由的风险。阿达·洛夫莱斯研究所的苏梅达·德什穆克告诉《连线》，独立专家应对这类预测进行测试，政府应衡量承诺的收益。
Dame Chi Onwurah chairs the Commons science and technology committee. She told WIRED that overstating what technology can deliver risks undermining the case for change. Sumedha Deshmukh of the Ada Lovelace Institute told WIRED that independent experts should test such forecasts, and that the government should measure the promised gains.
斯宾塞持续追问。11月4日，纳拉扬在回答他的一个追问时，指引回此前关于人工智能技能的一个回答。同一周，纳拉扬警告称，英国发生严重网络入侵“极有可能”。
Spencer has kept asking. On 4 November, Narayan answered a follow-up question from him by pointing back to an earlier answer on AI skills. The same week, Narayan warned that a serious cyber breach in the UK is “extremely likely”.