If there’s any truth left to the cliche that middle management adds little more than bureaucracy and communication overhead to an organization, then Slack’s manager bot product should fix that market inefficiency pretty quickly. Status reports and work broadcasts are ripe to be automated and the future points to an eventuality of AI-generated communications.
Algorithms are pretty good at encapsulating narrowly-defined tasks, and there’s nothing about software that restricts its ability to replace work for both low- and highly-skilled work. In fact, the “routine”-ness of job better captures whether it can be (and is being) automated, and AI is poised to take over non-routine work as well.
So I agree with the premise of speeding up routine tasks in general project management; where it breaks down is equivocating that narrow band of responsibilities to overall people management. Sure, there are some pretty bad managers out there who do reduce their work to little more than project pushing, but the better managers out there also worry about career advancement, mentorship, performance, team building, product strategy, and filling in whatever gaps need to be filled on the team.
And while there’s certainly the possibility of having individual systems — or even a grand overarching system — optimized for each of these areas, getting to that state is a huge stretch from where we are now. I can think of two major roadblocks to AI-as-manager:
- Many types of work are hard to quantify. There’s a lot of types of work, particularly in creative fields, where we have trouble measuring how well of a job someone is doing on a day-to-day basis1.
- People still prefer to interact with other people. This is cultural norm that may change as more and more machinery takes over aspects of life, but I suspect people management will be one of the last holdouts.
That said, AI to monitor work seems like a reasonable first step, and the most routine of management responsibilities to take over. If it’s useful, then at worst it’ll free up more time for other things, which still makes these reporting bots a helpful tool.
The corollary, though, is that we have found ways to quantify and thus partially automate work which ends up being routine. The best example I can think of is the assembly-line-like work of fast food kitchens and their eventual robotization.↩