Last reviewed: May 24, 2026

What is augmented automation? Definition and right instinct

Augmented automation is the use of AI to assist a human worker in performing a task, rather than to replace them. The human remains decision-maker and responsible, the AI accelerates, prepares, or validates. It is the dominant mode in 2026 for tasks with legal, creative, or relational stakes.

Augmented automation is opposed to full automation (replacement). Three characteristics define it. Human in the loop: at each critical step, a human validates, corrects, or rejects the AI's proposal. Traceability: each joint decision is logged, which allows audit and progress. Preserved responsibility: the final decision remains legally attributed to the human. In 2026, 35.9% of US workers used generative AI by December 2025 (Federal Reserve Bank of St. Louis), mostly in augmentation mode. Documented productivity gains are substantial: 15% on average, 30 to 35% for the least experienced workers (Brynjolfsson, Li, Raymond, study of 5,179 agents at a Fortune 500 firm published in Quarterly Journal of Economics 2025). Gains are concentrated on junior profiles.

Concrete example

An 80-employee law firm deploys augmented automation for the first drafting of legal conclusions. The AI proposes a draft from the client file, the junior lawyer corrects and completes (20 to 35% of text modified on average), the senior lawyer validates before filing. Three benefits measured over 6 months. Drafting time of a first draft: 4 hours against 11 hours previously. Quality of the final deliverable (measured by client feedback): stable. Effect on juniors: they report ramping up faster, validating Brynjolfsson's hypothesis that AI disseminates best practices from seniors to juniors.

Good vs bad reflex

BAD REFLEX

Presenting each AI project as pure automation aimed at eliminating positions. This approach systematically triggers three resistances: union and social opposition, refusal of cooperation from the operators concerned (who do everything to sabotage or minimise the tool), and legal risk on unexplainable automated decisions. The project fails at the deployment phase, regardless of the model's technical quality.

GOOD REFLEX

Positioning AI as augmentation of existing collaborators, by measuring the gains it brings them: time saved, improved quality, ramp-up of juniors. Augmentation rhetoric aligns interests (collaborators, executives, legal, compliance) and triggers faster adoption. Documented productivity gains (15% on average, 30 to 35% for juniors) are higher than gains from a poorly accepted full automation.

Sources

Generative AI at Work, Brynjolfsson, Li & Raymond, Quarterly Journal of Economics vol. 140 n° 2, 2025. https://academic.oup.com/qje/article/140/2/889/7990658 (accessed 2026-05-24)
The Rapid Adoption of Generative AI, Bick, Blandin & Deming, Federal Reserve Bank of St. Louis Working Paper 2024-027C, revised 2025. https://www.stlouisfed.org/on-the-economy/2025/feb/impact-generative-ai-work-productivity (accessed 2026-05-24)

← Back to glossary

What is augmented automation? Definition and right instinct

Concrete example

See also

Further reading

Sources