On AI, Late and Early

I’m hesitant to republish older content on LinkedIn — particularly around AI, because everyone and their uncle cosplays as a newly minted AI expert on this platform.

I’m not an AI expert. But 30-plus years in communications and persuasion offers a sense of the territory, if not an exact copy of the map. The path to today was probably clear to researchers before Google published its seminal Attention is All You Need paper in 2017. I was late to that party, but I thought through some of the ramifications 28 months before ChatGPT was a consumer product.

I could lean on that and call myself prescient. But instead I wonder: What’s the 2017-esque thing I’m missing now?

July 21, 2020

GPT-3 And The Future Of Us

In case you missed it: There’s been a lot of news around GPT-3, a machine-learning model rolled out recently that’s a full order of magnitude larger than its predecessors.

What does it do? Put simply: It writes.

It writes stories in the style of well-known authors on demand, poetry, news stories, etc. In tests, most people can’t tell the difference between GPT-3 news stories and those written by a human.

GPT-3 shows that it’s possible for a model to someday reach human levels of generalization. It’s not practical right now, but once the impossible becomes possible, it’s only a matter of time until it becomes practical.

A decade from now or maybe less, it’s not a stretch to see most low-level content jobs — from clickbait to news production — replaced with on-demand access to something like this. It’ll also eventually lead to changes in financial markets; trading based on the semantic parsing of the day’s global news feed seems like too big a prize to be ignored.

And, of course, trust will take another leg down as people no longer just have to worry about individual or institutional bias in what they read, but algorithmic bias as well.

July 23, 2020

Automating Attention

There’s a lot of hand-wringing over how AI will automate finance, medicine, or writing. But my guess is the real game-changer will be automating attention.

We already customize ads and content for individual people; think how much more attention could be captured with a digital avatar matched to our tastes and preferences.

I don’t worry much that digital models — with programmer-assigned backstories and favorite political or social causes to make them more relatable — are a basis for appropriation. But it’s an interesting thought. If a digital celebrity tells you to fight the power, but is a wholly owned creation of the power itself? That’s not too much of a stretch from where flesh-and-bone celeb culture is today.

One side effect: Digital properties can be repurposed at the drop of a hat. It’s not hard to imagine a future where the digital it girl/guy of last year’s Calvin Klein campaign isn’t hitting their numbers anymore, so it’s sold to B-list TV shows or porn producers.

Greg W. Brooks. First published on LinkedIn, April 5, 2026.