Future of Knowledge Work

Knowledge workers are broadly the professionals who use primarily their brain to do their job, as opposed to jobs that require physical exertion. Another way people refer to this distinction is white collar vs blue collar work, although it’s not quite the same because something like driving is primarily a mental activity that is considered blue collar.

For a while, the general belief was that knowledge work would be harder to automate, and therefore had an advantage as a choice of employment. This was substantiated by most of the automation we’ve seen since the industrial revolution, which generally was in the form of mechanical super powers. However, there is a concept, called Moravec’s paradox, which basically says that things that are easy for humans (or animals), like walking, are hard for machines; and things that are hard for humans, like arithmetic, are very easy for machines.

One possible explanation is that we’ve had millions of years to evolve the ability to move, whereas critical thinking is a much newer process and therefore evolution doesn’t have a long lead time. Regardless of the underlying principles, with advancements in AI over the last decade we are seeing a shift towards automation of knowledge work.

AI is close to being able to drive autonomously, read and write complex text, generate visual and musical creations, and become a personal assistant. Accounting, Law, Medicine, Management, DJ, are all professions that not long ago were considered immune to AI disruption, which are now ceding ground. At the same time there is more motivation to disrupt these professions because the hourly cost (and therefore upside) is higher.

How will AI change the way we work?

We’ve already experienced a huge shift in how we work in teams towards more collaboration. Thanks to software and the internet, and exacerbated by the lockdown, we now have more ways to collaborate than ever. Collaboration is necessary to accomplish progressively larger objectives, but it comes with overhead on communication and coordination. When individuals are not aligned, collaboration can be frustrating and costly.

The next phase is learning to collaborate with AIs. This will come with its own nuances, because for the foreseeable future AIs will not be as smart as humans. So we will need to learn the interface with AIs, like we learned the graphic user interface for photoshop. AI interfaces are going to be more “natural” in the sense that we will be able to use language, gestures, mock-ups, just like we communicate with humans. This will provide more flexibility and allow us to express highly complex ideas, but it will also make our learning more meandering and glitches harder to debug.

In a sense, collaborating with an AI is akin to being a manager. You need to know the strengths and weaknesses of the individual you’re managing, and you have to be able to break down a complex project among the members of the team.

Is AI a threat to the creative industry?

Without a doubt, I believe AI will continue to be capable of matching or beating humans at tasks we consider to be within the creative industry. This is unavoidable. The reaction should not be fear. Instead, creative professionals should learn to collaborate with AI as a tool, just like they learned to use software like Photoshop. For example, if an AI takes a sentence and generates a visual creative, a person still needs to play with different prompts and ultimately select the best output. Understanding the technology and its limitations will be key. The skills that make someone great today might change, just like they did when creative work first became digital.