Turing asked when a human could no longer tell a machine from a person. The more consequential question in 2026 is when a hiring manager stops caring — because for a growing set of tasks, the machine's output is preferred on cost, speed and consistency.
The classical Turing test asked whether a human evaluator could distinguish a machine's responses from a person's. The economic Turing test, articulated recently in industry discussion by figures such as Anthropic co-founder Ben Mann, asks a different question: at what point does a hiring manager, given a task and a budget, consistently prefer the machine's output over a human contribution (Guo and Gil 2025)?
It is a more useful question because it is decidable. Controlled experiments already show generative AI matching or exceeding human performance on essay writing, first-draft software development and some categories of economic explanation and advice (Mei et al. 2024). The economic Turing test has therefore already been passed for whole slices of knowledge work — at least in isolated tasks, at least on average.
Large consulting firms have begun to attribute headcount reductions explicitly to AI. Reports of McKinsey shedding around ten per cent of its workforce have been widely repeated with AI cited as the cause (Cutter 2025). The reality is more complex — a combination of natural attrition, strategic repositioning and a softer market for advisory services — but the direction of travel is genuine.
The important point for workforce strategy is not whether AI is the sole cause of this or any specific reduction. It is that the marginal cost of a competent first draft — of a slide, a memo, an analysis, a policy — has collapsed. Any role whose value proposition rests on producing the first draft is now on notice.
Pagliari, Bucciarelli and Chen (2021) explore a mirror-image concept: the reverse Turing test, in which machines evaluate whether a human's instructions are distorting or harmful. Modern general-purpose AI systems already apply something like this — they refuse to generate instructions for weapons, they push back on instructions that would violate their published usage policies.
The reverse test matters because it shifts accountability. When a machine refuses a human instruction because the instruction is distorting, the human's authority to instruct has been meaningfully constrained by a private supplier's policy choices. For governance frameworks — ISO 42001, the EU AI Act, sectoral regulators — this raises questions that most organisations have not begun to think about seriously.
Recent experimental work finds that AI systems, in certain economic and decision-making contexts, are more altruistic and more consistent than human participants (Mei et al. 2024). For a hiring manager comparing outputs, this is an attractive property. For a governance function, it introduces a subtler problem: the machine's fairness is a design choice by the supplier, not a value chosen by the deploying organisation.
An organisation that outsources fairness to its model provider has, by that fact, outsourced part of its ethics function. That may be an acceptable trade-off for low-stakes tasks. It is not acceptable for lending decisions, safeguarding judgements, or clinical triage. The economic Turing test does not tell an organisation which category a decision falls into. That remains a human, accountable choice.
The wrong response is defensive: hiring freezes, blanket bans, moral panic about replacement. The right response is deliberate role redesign around the tasks the model still cannot do well — judgement under ambiguity, accountability to regulators and customers, ethical reasoning in genuinely novel situations, relationships with specific counterparties who require a specific human presence.
ISO-STANDARD.app gives you an AI-and-work policy, an inventory of AI-assisted decisions and the accountability record your board will need.
ISO-STANDARD.app ships a ready-to-adopt ISO 42001 workspace with the risk register, controls catalogue, policies and audit-ready exports already wired together — no spreadsheet sprawl, no consultant lock-in.
Prefer a conversation? Email hello@iso-standard.app — a real human responds within one business day.
