From deliberative practice to algorithmic theatre

Generative AI is usually defended in the language of efficiency. Environmental due diligence, however, rests on documents machines can now draft, submit and summarise. This article proposes algorithmic theatre as an explanatory model of a risk in which compliance artefacts keep their legal form and authority while the deliberation they record thins or disappears. Using Dryzek’s criteria of authenticity, inclusiveness and consequentiality, it reads three episodes as plausibility probes (AI pilots under the National Environmental Policy Act, fabricated public comments and false data in carbon credit verification) and specifies three hypothesised mechanisms: de-skilling, decoupling and legitimacy transfer. A comparison of the United States, the European Union and Pakistan suggests that exposure depends less on a jurisdiction’s wealth than on whether its law attends to the provenance of compliance information, and is greatest where participatory infrastructure is already weak. Safeguards follow, including a draft provision for the disasters convention now under negotiation.