On productivity and AI. It's complicated.

Part 2 of my talk on the Acadian Ventures AI thesis.

Read the first part if you like here.

Productivity is a complicated concept. It is really hard to get to genuine causality with productivity improvements, and even defining what productivity actually is seems to keep economists up at night. If you would like to also feel perplexed, read about the Solow computer paradox.

It is very easy for shiny object syndrome to have us imbue technology with all sorts qualities it doesn’t actually have. Our imaginations are both powerful and dangerous.

This quote from Marc is powerful reminder that we need to look beyond the potential of a technology to actual deployment in the real world. The possibilities seem endless, but markets and organizations determine value, not the technology itself.

The economists can’t yet agree on the impact of AI on productivity. Daron Acemoglu is a good place to start. His prediction is pretty modest.

Other economists / computer scientists are more bullish on the productivity improvements. See Andrew McAfee for instance. He has a long career in academia, and has recently started a company to help organizations figure out AI’s productivity impact. He and his colleagues are doing some pioneering work on getting to causality measurement. There is still a long way to go, but its promising. Workhelix is one of the companies I wish we could have invested in, but hey, not every founder returns our calls.

When vendors or consultants suggest massive ROI numbers, I’d urge you to challenge them. Ask them how they got to the numbers. When every tiny little feature is heralded as a revolution, let’s call it out as hype and bullshit. I’m confident that we will see some genuine productivity gains, but proof of concepts aren’t proof.

The next section will look at AI’s impact on jobs. I’ll post that in the next couple of days.

In the meantime, I’ll leave you with a song. I watched the Bob Dylan movie on the plane the other day, it was really good. I was thinking of times are changin’, but when it comes to productivity and AI, it’s still blowing in the wind.