This post first appeared on LinkedIn.

I’m more excited about generative AI than any other information technology since the web — but not for the reasons you may think.

It’s not because AI will solve all intractable problems. It won’t. Some things (e.g., misinformation) may get worse in the near term.

It’s not because AI will lead to unprecedented economic growth. It might, but it might also aggravate inequality and environmental issues.

It’s not because AI is sentient. Spoiler: it isn’t. A few minutes working with even the most advanced image-generating models will disabuse you of that idea.

Most people are missing the point of gen AI, primarily because of how these technologies have been framed and marketed.

Generative models aren’t “an intelligence” in the sense most of us understand intelligence. Moreover, I’m skeptical that current approaches will lead to the type of AGI that eliminates all our jobs and inadvertently turns us into paperclips.

Instead, generative models are an astonishingly powerful symbol manipulation technology.

Think of what spreadsheets do for working with numbers. Spreadsheets give you superpowers: forecasting, modeling, analysis, data capture and transformation, etc. Critically, you don’t need to be a programmer to acquire these abilities.

But at their core, spreadsheets address a relatively limited domain: math. Fundamentally, spreadsheets are for processing numbers; they augment your ability to work with mathematical constructs.

That’s no small thing; numbers can take you far. But Gödel notwithstanding, math has fairly predictable rules, structures, and outcomes. 2 + 2 = 4 leaves little room for controversy.

Now envision a system that lets you do the types of things spreadsheets do with numbers, but that works with the entirety of humanity’s symbolic production.

Not just numbers, but every concept that’s ever been expressed or can be expressed – with all the breadth, flexibility, ambiguity, potential, and yes, risk, that implies.

That’s a big deal.

Shoutout to Greg Wilson, whose Mastodon post inspired these thoughts.