Will AI replace 'knowledge workers'?

I’ve been hesitant to wade into the AI cycle. But here we go.

Will AI replace “knowledge workers?”

Large language models, like GPT-4, learn from unstructured text. We could spin our wheels pondering to what extent such a model acquires the kinds of “knowledge” that “knowledge workers” “produce.” But none of this is about what machines can do, is it? It’s about what happens to us. Namely, to our jobs.

David Graeber wrote about bullshit jobs—jobs that serve chiefly to keep people working, which is to say, docile. It so happens that LLMs may replicate the work many of these bullshit jobs entail. But, as Graeber argues, these jobs never needed to exist in the first place.

What’s more likely: a secular trend toward unemployment across many knowledge-intensive sectors? Or that a new system will find a way to keep the same people docile?

It’s not clear yet. This uncertainty is what I think all the AI talk is really about. (AI x-risk serves as an allegory for this uncertainty in many people’s minds. After all, it’s easier to imagine the end of the world than the end of capitalism.)

Is there a third path, here, hiding, somewhere? An old-new story of liberation, for a people who don’t yet know how to ask for it?

Thanks to Zeke Medley for the conversation.