3 Comments

AI "shadow employees" seem pretty inevitable, right?

Pretty soon, it seems extremely likely that we'll have the tools to set up an AI to watch and listen to you while you work, and then support you by performing automated tasks (e.g., queries/analysis/research/etc. based on the emails you're getting, your to-do list, your calendar, etc.) To start they would just be supervised "suggestion bots" without authority to send emails or write files unless approved, but they'll still be able to automate a ton of work, and improve the output of their human counterpart.

On one hand, this will be AWESOME! If I could hand over half my workload to an AI, that would be amazing, especially if it handles the annoying stuff that disrupts your day like people sending you emails looking for routine information.

As long as you are in personal control of the AI, it sounds great, right? But what if your employer forces you to be constantly recorded by an AI that is learning your job? Obviously there's a privacy concern, as well as the fear of losing your job as soon as the AI approaches being as good at the job as you.

But, there could also be benefits to having a company with super-productive AIs - you would have access to so much more analytical power than if you rely on humans. And you know Hank in IT who always takes 5 days to respond to an email? Now his shadow AI can respond immediately and give you the info you need.

So personally, if I were in control of it I would love to have an AI automating my job. Even if my employer controlled it, I think I would still be OK with it, just because I feel secure in my ability to contribute, and if my job were able to be fully supplanted, that seems like we would already be living in a post-scarcity utopia.

Expand full comment
author
Jun 17, 2022·edited Jun 17, 2022Author

> I think I would still be OK with it, just because I feel secure in my ability to contribute

That's the important bit on this discussion, I think. It's not clear if we would still be able to contribute, and that became patent in the case of Tim. It's an open question if the augmentation promoted by AI will open new venues for human ingenuity or if it will concentrate all the work in a few high throughput individuals.

But let's say that yes, Tim today received his letter of termination because his work can't compete with an AI. He is a smart guy, though, so he looks at the market and searches for a job where his abilities can be of good use for society. Maybe he chose to learn to code and landed on a software engineer position. But how safe can he really feel that once again his craft won't be better served by an AI?

I think workers are set to lose a lot of leverage in the next decades. The process of substitution will be gradual, but inevitable. Unless our demand for code / art / whatever increases in the same magnitude as technology, we will see the relative value of work force decreasing. There will be some winners, though: the few ones that will be lucky enough to be responsible for managing all this AI productivity surplus. But they will be a minority.

Expand full comment

Even if programs like Dalle-2 generate variable quality output, that could turn into a scenario where the audience becomes the co-creator. It would be easy to create feedback mechanisms from audience impressions back to the creative engine, so there wouldn't need to be a single "creative human" in the loop making decisions. This feedback loop could generate frighteningly compelling content that appeals to the lowest common denominator of human engagement (not that different to the dynamic created social media feedback mechanisms).

Expand full comment