I've put a date on a bet about AI agents and Cybernetics
Firms deploying AI agents are running into problems Ashbyan Cybernetics described decades ago. I have written down what I expect to happen next, dated it, and will keep the evidence in public, including the evidence against me.
Richard S O'Rourke
If you work with AI agents, you have probably met some of these already: more variety than any control can match, goals that drift, work that nobody independently checks, and an organisation that cannot absorb what its agents change. None of them is new. Ashbyan Cybernetics described them decades ago, and the work of Ashby, Beer and Pask is what my research rests on.
On 6 October 2026 I made a dated prediction about what happens next. It is now on this site as A bet on Cybernetics and AI agents.
The bet, in two parts¶
The first part is that the problems have returned. AI agents, and the organisations that deploy them, are already running into the problems Ashbyan Cybernetics described. Firms will name and solve these problems, most likely under new names.
The second part is that the old solutions will get a fresh airing. I expect 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. I measure it as citations per million papers in each field, against the years 2015 to 2022.
I may lose the second part. At least three times before, fields building machines that adapt met these problems and started fresh: control theory in the 1950s, autonomous driving after the DARPA challenges, and the multi-agent systems of 1995 to 2010. So far, citations of Ashby's An Introduction to Cybernetics per million papers in each field have been flat or falling since 2022. Three things are different this time. The agents are being deployed inside organisations, where Beer built Ashbyan management Cybernetics. They work in language and conversation, which is Pask's ground. And the word is returning in prominent places.
Why make a public, dated bet¶
Because it can be checked. Anyone can say that an old idea matters again. A dated claim, with a stated measure and a record kept in the open, can turn out wrong, and I have said in advance how it might.
The first part is the one that matters to the people deploying agents. It is about the problems they face now, not about which books get cited. If it 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. I apply this research at Intelligence Amplified, where How much should you check the AI's work? shows one of the problems at its simplest.
How I track it¶
Each week I scan the literature for new work that cites the founders of Ashbyan Cybernetics or uses the word “cybernetics”. I read the papers that matter, classify the sentences in which they cite the founders, and publish the best of them on the bet page with my comment. Charts for each author, and answers to a few standing questions, follow as the record grows.
The evidence is already arriving from both directions. Since May 2026 at least five papers on AI agents and their governance have built on Ashby by name. A September paper on automating AI research, by twenty-two authors including Hinton and Bengio, described its own mechanism as “a recursive feedback loop” and cites neither Wiener nor Ashby.
Follow it¶
The bet page is where the record lives, and I update it as the evidence comes in.