人工智能劳动力：从炒作到硬核事实

What it takes to deliver AI value KPMG. Make the Difference. Therushisover.Nowit’stimetomakeAIwork This report outlines our perspective ondelivering real AI value, drawing on four yearsof experience working with C-suite leaders andobserving how organisations adopt AI acrosstheir workforces. It highlights the critical lessonslearned in translating technology investment intomeasurable impact and explores how emergingcapabilities - like agentic AI - are reshapingroles, workflows, and the skills people needto succeed. By combining insights from bothtechnology and human behaviour, we providea roadmap for organisations seeking to movebeyond experimentation and build sustainableAI-driven performance. Over this period, our understanding andapproaches have also significantly changed. AIrequires a completely different approach thantraditional transformation, one that is ironicallymore focused on people than the technology. Billions have been spent chasing the AIwave - but too often, that money wentinto licenses without logic, pilots withoutpurpose, and hype without a plan.Thereality? Most organisations are nowhere nearcapturing AI’s full potential. Why? Becausethey treated AI like a system upgrade, not astrategic revolution. AI: a people and business revolution One of the challenges corporates are facing innavigating the AI revolution is the lack of clarity associety at large still struggles to understand theimpact of the technology. Big questions remain,such as: How do you regulate effectively withoutdisadvantaging your market? How do you preventAI capability falling into the hands of individualswith the capacity to harm? And what does thefuture of work look like when AI can perform alarge percentage of work done today? Why AI value remains elusive Over the last 3 years, we have studied business’sreactions to AI technology and the journeymany organisations have been on seeking the“promised benefits that AI will deliver”. Whilstthere has been success in some areas, themajority still have a way to go before translatingtheir investment into true business value. xx We’re still at the early stage of the adoptioncurve in the workplace — innovators andearly adopters are pushing boundaries,but mass adoption is yet to come. Whilst the answers to the above are still indevelopment, one thing is for certain, AI is goingto dramatically change the way in which work iscompleted and will require a measured approach,bringing everyone on the journey. When AI truly took its place at the centre oftraditional business operations in 2022, bothbusinesses and consultancies started trying tounderstand what this meant for them. Manybegan to experiment and get ‘hands on’ with thetooling, whilst others stood back to see how thefirst movers faired in the new world. Michael Allen - People Consulting Partner Contents 14 04 At a glance From hype to impact: evolving yourAI strategy for sustainable value18 Lessons from the last year 09 How KPMG can help Terms used throught this report Adoption–The process of integrating AI into everyday business operations.AI–Technology that enables machines to mimic human intelligence.AITooling– AI enabled tools which supports employees doing current tasks more effectively. E.g. Microsoft CopilotAgents/Agentic AI– AI agents complete processes and are able to make decisions. Agents are a newer form of AIwhich supports the reimagining of processes.LLMs (Large Language Models)– Advanced AI models capable of understanding and generatinghuman-like text. From hype to impact: evolving yourAI strategy for sustainable value Evolving systems demandevolving people Ataglance Yet, trust remains a critical challenge.Globally, only 46% of people arewilling to trust AI systems (Trust,attitudes and use of artificialintelligence: A global study,2025). This makes transparency,governance, and user confidenceessential for organisations seekingto realise the full value of theirAI investments. AI has moved rapidly from hype tohard truths. Organisations across everysector have invested heavily in new toolsand pilots, but too often the return hasbeen modest. Many organisationshave yet to realisevalue from AI1 Progress variessignificantly acrosssectors2 Sectors such as retail have movedfast, often prioritising cost savingsand quick ROI, while manufacturingand energy remain cautious,slowed by scale and complexity.Healthcare shows ambitionbut is fragmented in execution,while banking and insurance faceregulatory hurdles that createtension between opportunity andrisk. These differences highlightthat there is no universal blueprint– strategies must reflect sectordynamics, competitive pressures,and regulatory context. Despite bold investments, toofew programmes have deliveredsustained returns. Lessons fromour clients show that failureoften stems from unclear valuetracking, applying old ways ofworking to new technology, orrolling out capability withoutunderstanding how AI fits intoroles and workflows. Withoutproof