AI at Work from Productivity Hacks to Organizational Transformation 2026

Despite early signals, three fundamental questions remain unresolved. How organizations and policy- makers address these uncertainties will determine whether AI creates broad prosperity or concentrates power and risk. 1 What is “higher-value work” in practice? Executives frequently shared the observation that AI handled workers’ routine tasks, which allowed humans to focus on “higher-value work”. But what is that work, specifically? Without clear definitions, “higher-value work” risks becoming an empty promise – or worse, a way to veil displacement. The critical questions include: is there enough genuinely high-value work to absorb workers whose tasks AI automates? How do organizations create new categories of valuable work? Does “higher value” mean strategic thinking, creative problem-solving, spending more time with clients or something else entirely? If organizations cannot articulate what higher-value work means in practice, AI adoption may deliver productivity without broadly shared prosperity. This vagueness reflects genuine uncertainty about what humans will do in AI-augmented workplaces. 2 What does organizational redesign really mean? Companies talked extensively about adopting “AI at scale” and embracing “AI-native management”, but concrete models remain elusive. Executives suggested that workplaces would have flatter hierarchies, faster decision cycles and more adaptive workflows. Yet few could articulate concretely what this looks like in practice or how it differs between contexts. The uncertainty extends across multiple dimensions. A 50-person start-up will implement AI differently from a 100,000-employee enterprise, but the principles distinguishing these approaches remain unclear. When firms describe multi-agent systems managing entire workflows, they may be imagining leaner organizations with fewer management layers. But these redesigns may simply add new complexity without reducing the old. The vocabulary for describing these transformations and modelling their impacts does not yet exist. The challenges are both design- and workforce- related: if the traditional roles of mid-tier coordinators change or compress, what replaces them? Companies are moving from pilots to systematic integration without clear templates for what success looks like. Without clearer models, executives risk replicating structures that might work well for tech giants but fail in healthcare, manufacturing or public-sector contexts where the constraints and objectives differ fundamentally. 3 Who is responsible for decisions made by AI? As AI systems make increasingly consequential decisions – such as approving loans and diagnosing health conditions – the question of accountability becomes urgent but remains largely unanswered. Who owns the outcomes when AI- mediated systems produce answers that no one wants but also no one ultimately controls? Economist Dan Davies calls these problems “accountability sinks” – like when a gate agent cannot explain an automatic rebooking after a flight mishap, or an insurance agent is powerless to override a claims denial.13 These frustrations already permeate modern organizations. AI threatens to multiply and compound them. Similarly, “agentic identity” is an emerging area within cybersecurity dealing with technical, legal and governance questions that companies adopting agentic AI will need to navigate. Does an AI agent have the same identity and permissions as its creator, or does it have its own? Who is liable when an AI agent makes an error: the vendor that supplied the AI agent or the customer that ran it in their environment? The stakes of these questions extend beyond organizational efficiency and affect the legitimacy of institutions and people’s trust in governance. Without resolution, AI could create a world in which consequential decisions feel imposed rather than chosen, eroding public trust in institutions. The challenge is not just technical but organizational and political: building systems that preserve meaningful human agency even as they become more powerful.3.2 Three macro questions AI at Work: From Productivity Hacks to Organizational Transformation 19