While it’s perhaps fair to say that some of the considerable hype around AI and its ability to transform the world has died down a little in recent years, the expectations remain high that it will eventually get there.

As with many previous technologies, there is a strong sense that we have spent much of the past few years trying to replicate what has gone before (a faster horse if you will) rather than redesigning processes and systems so that they take advantage of what the technology can do.

In their latest book Power and Prediction, Ajay Agrawal, Avi Goldfarb, and Joshua Gans highlight this by outlining the three core ways in which a technology can be used.

• A point solution, which is when an existing procedure is improved upon. It can be adopted independently and doesn’t require the system within which it’s embedded to be changed.
• An application solution, which is when a new procedure is created and adopted independently. This also doesn’t require any changes to the system in which it’s embedded.
• A system solution, which is when existing procedures are improved, or new procedures created, by changing dependent procedures.

“The true transformation will only come when innovators turn their attention to creating system solutions,” they explain. “Those solutions themselves bring AI to an economywide scale, and their momentum spurs further application solutions.”

Suffice it to say, this thinking is not new, and indeed the need for business process re-engineering was underlined at the start of the computer age to explain why computers weren’t filtering through to the productivity figures. It’s nonetheless worth reminding ourselves that meaningful change from AI won’t come until we stop applying it to tasks that we’ve always performed ourselves.