Knowledge Work in the AI Era
Given the convergence of several trends, including AI and RTO mandates, I thought it worth revisiting Peter Drucker’s thoughts on knowledge worker productivity.
Drucker first coined the term "knowledge worker” back in 1966. At the close of the 20th century, he said that "the most valuable asset of a 21st-century institution, whether business or non-business, will be its knowledge workers and their productivity."
He identified six factors of knowledge worker productivity, all of which are still relevant 25 years on.
So let's take a look at Drucker's six factors of knowledge worker productivity, starting with...
Defining the Task
Back in 1999, Drucker wrote, “Knowledge worker productivity demands that we ask the question: 'What is the task?'"
My 21st century reflection?
You cannot automate a workflow without understanding the job to be done and what "good" looks like. There's a lot of FOMO around new tools (and I get three cold emails a day pitching AI products), but when it comes to replicating a desired outcome, you've got to nail it before you can scale it. Unless you figure out the process that will produce the results you want, you are automating crap inside a black box.
2. Worker Autonomy
The second productivity factor Drucker identified was autonomy, or in his words, “It demands that we impose the responsibility for their productivity on the individual knowledge workers themselves."
My thoughts:
The RTO command and control culture is squarely at odds with worker autonomy. It's hard to feel like you're responsible for your own productivity when you're told where, when, and how to work. In-person meetings can be productive and energizing, but if your team is distributed, why is leadership forcing people to commute into an office only to sit through back-to-back Zoom or Teams calls?
It seemed as though we were making progress through the forcing function of the pandemic, but now we're back to ignoring the power a sense of ownership instills in skilled workers. Why do companies insist on hiring smart people and then managing them like interns?
Maybe someone has already built a DruckerBot to dispense management wisdom on command to help leaderships navigate the current workplace reality.
3. Continuing Innovation
"Continuing innovation has to be part of the work, the task, and the responsibility of the knowledge workers."
I'm guessing Drucker would have encouraged knowledge workers to figure out how to incorporate AI into their existing workflows or expand the bounds of what's possible. But I don't think he could have anticipated how often the LLM foundational models would be updated or overhauled, or how tough it would be to keep up.
It can also be challenging to manage leadership expectations of AI innovation when not all of us can code, and have to work within the bounds of our company's tech stack and data infrastructure. For example, I started investigating how to analyze aggregated call transcripts for win/loss analysis, and even setting up something to test has proved arduous without knowing how to write APIs or having immediate access to someone to assist. There seems to be an attitude of "get AI to do it" when we encounter a thorny problem, without considering the specific steps we are trying to automate, and what (still mostly human) resources are needed to test and refine.
Business leaders, if you want your teams to innovate, you've got to put the structures in place. Otherwise, you are bogging them down with orgslop.
4. Teaching and Learning
Teaching is an important and related component of learning. It's why many learning projects end with students presenting what they've learned to others. Peter Drucker considered this balance of teaching and learning a factor of knowledge worker productivity:
"Knowledge work requires continuous learning on the part of the knowledge worker, but equally continuous teaching on the part of the knowledge worker."
When I consider the current work landscape, I wonder, 𝐖𝐡𝐨𝐦 𝐝𝐨 𝐰𝐞 𝐭𝐞𝐚𝐜𝐡 𝐢𝐟 𝐰𝐞 𝐝𝐨𝐧'𝐭 𝐡𝐚𝐯𝐞 𝐞𝐧𝐭𝐫𝐲-𝐥𝐞𝐯𝐞𝐥 𝐬𝐭𝐚𝐟𝐟 𝐚𝐧𝐲𝐦𝐨𝐫𝐞?
Can you get better at your craft when you are training an entity that doesn't learn in the human sense?
And are people really invested in teaching anymore, or are they hoarding their knowledge and doling it out as a form of self-promotion? (As they say on LinkedIn, "Comment for my proven framework.")
Furthermore, it seems as though fewer and fewer employers are investing in developing their staff or offering career pathways. Which is no doubt why there are so many knowledge workers on LinkedIn or in virtual professional communities, seeking to learn from others.
Do we need to bring back guilds?
5. The Importance of Quality
As Peter Drucker put it back in the '90s, "productivity of the knowledge worker is not...a matter of the quantity of output. Quality is at least as important."
However, dive into articles and social media posts about AI use in the workplace, and you’ll see they focus on volume rather than efficacy as the primary gain. Perhaps this is because it's too early to measure the results of generative AI, but emphasizing volume without regard to impact sounds like a real drag on productivity.
What's the conversion rate on the marketing content churned out with an LLM, versus the content you crafted by a professional? What is the balance of human vs. AI-generated output?
When you're forced to do more with less, what is the quality bar for "more"?
If we are all prioritizing "volume" as a measure of productivity, are we even knowledge workers anymore, or are we on the factory line?
6. Knowledge Workers Are Assets
“Finally, knowledge worker productivity requires that the knowledge worker is seen and treated as an ‘asset’ rather than a ‘cost.”
People are assets; AI is a cost.
That's the mindset necessary for knowledge worker productivity.
But right now it doesn't seem as though corporate executives are treating knowledge workers as assets--just look at the waves of tech layoffs attributed to AI. It seems as though, instead of valuable assets, knowledge workers are being treated as expendable seat warmers whose utility expires the minute they train a digital replacement.
This is a pretty dismal view of human potential, especially when LLMs don't exactly "know" anything.
So why are CEOs so eager to replace the people who do the real knowledge work? Because they've ceased to see human capital as an asset. They've either confused the worker with the tool, or they are selling a story to justify their investments and product strategy.
Either way, their remaining employees will not be as productive as they could be.