The mismatch between AI’s actual utility and its hype reminded me of Prediction Machines[1], which frames technological change as progressing from point solutions → platform solutions → system solutions.
We’re still in the “what the heck is the point solution here” phase, with a lot of anticipation for platform and system-level shifts. There are some point solutions—like coding assistants—that make existing workflows more efficient and higher quality, but they haven’t translated easily to other domains. Platform solutions require completely rethinking workflows holistically, and system solutions demand restructuring everything that depends on those workflows. That’s going to be slow and messy. Including financially messy.
The book likens this to the introduction of electricity. Initially, electrification meant new individual machines in factories organized around steam power. Steam power was hard to turn on and off and not at all portable. Actually getting the full benefit of electricity meant redesigning factories around electricity use as-needed (not just when the steam engine was running) and spatially organize around task efficiency (not proximity to the steam energy production). All that was not a quick shift.
I very much sympathize with the author's frustration over hype that fails to understand the underlying technology and puts unwarranted faith in a small collection of corporate leaders. But I do think that this technology does have a high degree systems change potential and possibly the momentum to see it through this time. Not that we know how that will play out of which actors or forces will bring it to fruition. It really doesn't feel the same as the other tech crazes of the last two decades.
We’re still in the “what the heck is the point solution here” phase, with a lot of anticipation for platform and system-level shifts. There are some point solutions—like coding assistants—that make existing workflows more efficient and higher quality, but they haven’t translated easily to other domains. Platform solutions require completely rethinking workflows holistically, and system solutions demand restructuring everything that depends on those workflows. That’s going to be slow and messy. Including financially messy.
The book likens this to the introduction of electricity. Initially, electrification meant new individual machines in factories organized around steam power. Steam power was hard to turn on and off and not at all portable. Actually getting the full benefit of electricity meant redesigning factories around electricity use as-needed (not just when the steam engine was running) and spatially organize around task efficiency (not proximity to the steam energy production). All that was not a quick shift.
I very much sympathize with the author's frustration over hype that fails to understand the underlying technology and puts unwarranted faith in a small collection of corporate leaders. But I do think that this technology does have a high degree systems change potential and possibly the momentum to see it through this time. Not that we know how that will play out of which actors or forces will bring it to fruition. It really doesn't feel the same as the other tech crazes of the last two decades.
[1] https://www.predictionmachines.ai/