The hidden work caused by AI

For as long as automation and artificial intelligence has been around, you will have seen a two-sided debate. You’ve heard that AI is taking jobs. You’ve also heard, rightly, that AI creates new jobs.

But what you probably haven’t heard about is the hidden work caused by AI. That is, the work that isn’t advertised in a formal position; the work that gets taken on and absorbed into existing job titles.

Indeed, AI hides much of the work it creates — but that doesn’t mean it’s not there.

AI and the work it creates

Artificial intelligence, or AI, is an umbrella term for a host of technologies that aim to simulate human intelligence. It includes innovations such as machine learning, natural language processing, voice recognition, image classification, and so on.

There’s plenty of work created and fuelled by AI that we know about. For instance, the creation and coding of such software in the first place. The work of the IT team to deploy and maintain these AI tools. And indeed, the work of taking what the AI software offers and using it in a meaningful way.

In short, we know that AI creates a wealth of jobs spanning from development, to deployment, to ongoing data analysis and decision-making.

But we need to understand what jobs AI causes not just overtly, but covertly, too. Without doing so, we hobble the progress in ensuring ethical AI and avoiding unnecessary fear of job erasure.

So, what covert jobs is AI creating?

Hidden work

The first category of hidden work caused by AI revolves around the integration of AI-powered technologies into the workplace.

This is all about the work that arises from the initial disruption of adjusting to AI in the office. Specifically, the new way of doing things now that AI technology is available. Think of it this way: it’s not just the central IT team that deploys AI technologies. AI also creates extra (initial) work for the teams that will ultimately use it.

This work includes finding and utilising ways to make the AI technology useful and effective once it’s in place. It’s the work involved in learning how to best incorporate the AI-powered tools at play.

In short, there’s a considerable amount of hidden work involved in weaving AI technology into existing work practices, policies, and politics.

The workers behind the scenes of AI creation

It takes vast volumes of data to create artificial intelligence-powered programs. And much of that data needs to be labelled.

This marks another category of the hidden work caused by AI — the behind-the-scenes administrative workers of AI creation. (As opposed to the programmers, who generally receive the acknowledgement.)

This is the overlooked work that includes, for example:

  • Labelling images with tags for machine learning and image recognition algorithms
  • Transcribing audio for speech recognition technologies
  • Labelling data and content that depicts fraud, misinformation, or even violence, to teach filtering algorithms to recognise it  

The humans behind the curtain

The final category of hidden work caused by AI is generated by its limitations and its hype.

First, AI limitations create a certain workplace gap-filling. That is, completing the half-tasks and extra steps that need doing because AI needs it or causes the need for it, but cannot complete the work itself.  

Next, the work caused by AI hype is hidden by design. It’s the work people end up doing for unscrupulous companies that have products pretending to be AI. In reality, approximately 40% of European “AI start-ups” don’t use AI. They label their products as such in a snake oil marketing ploy. Behind the scenes, humans are doing the work – not AI.

Is AI taking jobs, or hiding them?

An AI-powered program, on its own, won’t unlock elevated efficiency and profitability. AI needs the hidden work of careful implementation and integration, constant training and monitoring, and ongoing adaptation. Without this hidden work, it’s just a shiny tool that’s not being used to its potential.

It’s important to shine a light on the hidden work caused by AI because that hidden work is, itself, so important. By being aware of it, developers can work with the hidden workers. And that, in turn, stands to not only help the growth of AI but also its acceptance.

Before automation, before smart assistants, was Emma Nutt

Are AI ethics impossible?

Why we should embrace automation disruption