Discussion about this post

User's avatar
Dave Friedman's avatar

Your penultimate paragraph is, I think, the critical piece. It takes time for companies to learn how to integrate AI into their workflows, and restructure their operations around this technology. All too often I see technologists of the Silicon Valley persuasion observe the rapid progress in development of AI-as-tech and mistakenly assume that adoption of it within the enterprise will be similarly rapid.

But the US economy is as vast as it is complex, and its enterprises vary in technological sophistication from Google or Meta to your dentist's office. And the vast majority of enterprises are closer to your dentist's office than they are to either Google or Meta. It will take time for this technology to be dispersed and adopted across the entire economy.

Expand full comment
J.K. Lund's avatar

Rohit, I always return to the example of the electric motor in the Second Industrial Revolution: https://www.lianeon.org/p/life-in-the-singularity-part-2

Electric motors began to replace steam engines and were better in just about every way, more power dense, more efficient, quieter, better torque…etc. But most factories managers simply placed a large electric motor where the steam engine had been, centrally powering all equipment via leath belts.

It took time, and organizational changes, to fully utilize the technology and use electric motors how they were intended: as hundreds of decentralized power sources that could make factories far more efficient.

I suspect that, to some extent, this is happening with AI. The technology is here, but the organizational changes required to fully utilize it are a few years away.

Expand full comment
14 more comments...

No posts