Our bet on AI agents
A dated prediction, made on 6 October 2026. We keep the evidence here as it arrives, including the evidence against us.
The bet, in two parts
1. The problems will return
AI agents, and the organisations that deploy them, will run into problems that cybernetics described decades ago: more variety than any control can match, goals that drift, work that nobody independently checks, and organisations that cannot absorb what their agents change. Firms will name and solve these problems, mostly under new names.
2. The names may not
We also expect the research our practice rests on, the work of Ashby, Beer and Pask, to be cited more often in business, philosophy, engineering and computer science: from 2024, and more strongly from 2027 as agents spread. We measure it as citations per million papers in each field, against the years 2015 to 2022.
Why we may lose the second part
Twice before, a field building autonomous machines met these problems and did not return to their source. Autonomous driving, after the DARPA challenges of 2004 to 2007, met the variety of the road head on and cited neither cybernetics nor the engineering tradition of intelligent control that had formalised it. The multi-agent systems of 1995 to 2010 did cite Ashby, more than AI has at any time since, and then moved on. Today's agent protocols do the work the FIPA standards did a generation ago, and their specifications do not mention FIPA.
Three things are different this time. The agents are being deployed inside organisations, where Beer built management cybernetics. They work in language and conversation, which is Pask's ground. And the word is returning in prominent places: in Science this year, Evans, Bratton and Agüera y Arcas called for governance “in the cybernetic sense of the term”.
The evidence so far
At 9 October 2026 the first part has early evidence and the second has none.
In September, a paper on automating AI research, by Chan, Mindermann and twenty co-authors including Hinton and Bengio, traced the intelligence explosion to I. J. Good, a member of the Ratio Club, and described its own mechanism as “a recursive feedback loop”. Its reference list cites neither Wiener nor Ashby. Read the paper.
Citations of Ashby's An Introduction to Cybernetics, per million papers in each of the four fields, have been flat or falling since 2022. Across all fields, papers citing the book rose from 166 in 2024 to 230 by October 2026, counting each book once.
Why it matters to a board
If the first part holds, the firms that recognise these problems early will spend less on agents that fail, and will know what to check before an agent's output reaches a client. That is the work we do. How much should you check the AI's work? shows one of the problems at its simplest.
How we track it
Each week we scan the literature for new work that cites the founders of cybernetics or uses the word. We read the papers that matter, classify the sentences in which they cite the founders, and publish the best of them here with our comment. Charts for each author, and answers to a few standing questions, follow as the record grows. The theory behind the bet is on rsorourke.com.