In the third installment of his series on AI, economist Marty Hart-Landsberg argues that layoffs attributed to agents are often part of a narrative aimed at investors.
Published on August 13, 2026 on his blog Reports from the Economic Front, the third part of Marty Hart-Landsberg's series attacks IA agents. The author begins by describing them without emphasis: software systems capable of dialogue with other software and external bases, of decomposing a complex set of tasks into ordered tasks and of chaining decisions without human revival — but only based on large language models.
A market that has become very crowded
It traces the 2026 shift: after Claude Code and Codex on the programming agents side, publishers started selling agents preconfigured by sector — legal, financial services, e-commerce, customer relationship, recruitment. Banks like Citi buy simultaneously from Anthropic, Google, Microsoft and OpenAI, these agents having different strengths and can combine. There is no shortage of promotional figures: Gartner projects that 15% of daily work decisions will be made independently by 2028, McKinsey argues that agents could already cover 44% of US work processes, and Microsoft AI boss Mustafa Suleyman announced in early 2026 the complete automation of most office jobs within two years.
The counter-argument
Criticism is about these stories. Hart-Landsberg relies on press and market analyses showing that a significant proportion of the redundancies presented as AI-related productivity gains mainly coincides with trade difficulties, past over-recruitment or loss of market share. Invoking AI then offers a strategic dressing to ordinary cups, in a context where Wall Street rewards this speech. The political thesis of the experiment stems from this observation: maintaining the fear of an apocalypse of employment feeds the income of the sector while undermining the confidence of employees in collective action.
Read the full article on the original source
Go to the source →