Matthew Clifford says: “There are no AI-shaped holes lying around”. That is how he reconciles "the facts that (a) AI is already powerful and (b) it’s having relatively little impact so far Making AI work today requires ripping up workflows and rebuilding for AI. This is hard and painful to do…"

Organisations look at AI and think surely we can make massive use of this either (on the good side) to do new things and solve hard problems for the benefit of all, and (on the bad side) simply to cut whole swathes of the workforce.

Beyond specific technical tasks, it can be daunting to identify where and how to apply AI effectively across an organisation. How would you rewire an entire department’s functions for AI?

I think that's why we're going to see a trend to simply treat ordinary human job profiles as those AI-shaped holes. Thinking in terms of roles rather than tasks or functions.

Imagine a virtual department of human-sized AIs, each with memory, communication, and even 'personality,' operating within existing channels and hierarchies. Managing an organisation becomes simpler if we think in terms of virtual people in recognisable roles, rather than an opaque system of tasks. As agent-based AIs advance, maintaining 'explainability' is crucial.

You can imagine a person-sized AI at Company X emailing a corresponding AI at company Y, or a or example, a person-sized AI at Company X could email a counterpart at Company Y, or a human, about a specific issue. Externally, it’s easier to engage with an organisation if you can address a particular role, regardless of whether it’s filled by a human or an AI.

Whether this means a hard-pressed workforce getting rows and rows of additional workers to solve problems and meet needs more effectively or whether it means 90% of staff being made redundant and replaced by person-sized AIs is not yet clear though I fear it will be the latter.

To be clear I have no particular enthusiasm for this kind of development because I don't trust capitalism with this technology. But we still have to learn how to think about it and understand it and make use of it as best we can.