AI applications should focus on collaboration over automation

Rather late, I came across this really nice piece in The Atlantic by Google’s James Manyika and economist David Autor about the tech industry’s fixation with replacing people with AI.

Here’s the crux of it:

We should insist on AI that can collaborate with doctors, teachers, lawyers, building contractors, and others—instead of AI that aims to automate them out of a job.

Their argument is that building systems designed for end-to-end job automation:

1. Understates the complexity of the interactions and contextual awareness that most jobs require
2. Discounts the importance of intrinsically human qualities like ethics and accountability in decision-making processes
3. Assumes that incomplete automation now will pave the way for complete automation down the line

I won’t try any harder to summarize their argument, which

Not to say that radical automation isn’t on the horizon. It probably is and we should be prepared for that. But the best opportunity in the near term isn’t so much automation as collaboration: systems that automate tasks which AI can perform reliably so that people can focus on higher value work.

We should be psyched

Manyika and Autor’s article resonated with me for two reasons. The first is because I’m really pscyhed about the potential of AI to improve people’s working lives. All jobs, no matter how creative or intellectually demanding, involve performing at least some tasks that are boring and repetitive. The more AI can automate those away, the more we can all focus on more rewarding, higher-value work.

We should all be excited about the prospect of our jobs sucking less. But we’re not. Instead, we’re worried about being made obselete. And with good reason! The news is full of scare hype about

I don’t want to make broad prognostications about the impact of AI on the future of work, a subject of extensive and ongoing research by people smarter than me.

But I would like to point out that this anxiety-driven narrative is largely fueled by the hubris of Silicon Valley fundraisers, who have every incentive to be as loud and absolutist about the future of AI as possible.

Their AI-powered app (meaning it makes API calls to ChatGPT) is absolutely going to wipe out software engineering and replace lawyers, accountants, designers, artists, and writers—pretty much everybody except VCs, startup founders, and consultants—with personalized AI assistants.

In the meantime, your new job is to write prompts, unless you have an actual skill like woodworking or plumbing, which will be a nice career until robots figure that out too.

Back to shop class?

This dystopian scenario might even sound plausible for a moment, but only if you assume that everything else about your life and work is going to stay the same. But why would it? Why should we expect a world in which professionals are suddenly much more productive, with much more time to spend on creative, higher-value work, to look so much like this one that bots can simply automate everything they do away?

The problem with this techno-fatalistic “AI is coming for your white-collar job” venture-capital fundraising fantasy isn’t just that it overestimates what AI can do in the short term. It’s that it lacks the imagination to consider how professions can evolve and what people can create when they have more time and freedom.

All of this is to say: instead of preaching doom for entire professions, the real opportunity lies in applying AI—with humility and deliberation—to the specific frictions within each field, so that all of us can spend more time doing what we love.

The views and opinions expressed in this post are my own and do not necessarily reflect those of the International Labour Organization (ILO) or its member States. Any errors or omissions are entirely my responsibility.