It’s a question I’ve been thinking about a lot. A few months ago, I was looking for a way to find AI use cases for ANY business. For weeks, I kept coming back to my list. (“Writing, summarizing, data analysis, etc...”) Every time, something felt off, and I couldn't figure out why.
And then that nagging feeling finally became clear:
I was trying to answer the wrong question.
“What tasks can generative AI do?” was the WRONG question. Or rather, it was the SECOND question.
The FIRST question we need to answer is: What do knowledge workers do?
Forget industries. Forget titles. Forget departments. When we talk about knowledge work, when we say AI will impact millions of knowledge workers, what exactly do we mean? What exactly do these knowledge workers do?
Answer *that*, and then “What can AI do?” starts to become a lot more clear.
“Very often people in companies are saying, ‘OK, how can I do this use-case? What can I do with gen AI?’ And I think this is a wrong question. The question they should ask themselves is, how can gen AI support me in solving a business question?”
— François Candelon, managing director and senior partner at Boston Consulting Group, WSJ
So, What Is Knowledge Work?
“It is information that enables knowledge workers to do their job.” — That’s according to Peter Drucker, who first coined the term in 1959.
Knowledge workers think for a living; they deal in information, and their output is knowledge. All tasks that knowledge workers perform can be grouped into one of eight categories, generally based on what the worker does with data. (That’s data in the broadest sense — information of any kind, whether it’s words in a document, numbers in a spreadsheet, or pixels in an image.)
They are:
- Research: Find or collect data on a topic.
- Learn: Understand a new topic or develop a new skill.
- Analyze: Interpret or find patterns in data (including summarization).
- Decide: Make a choice or recommendation based on data.
- Innovate: Remix or reimagine ideas in a new way.
- Create: Make something new (text, audio, video, image, code).
- Execute: Take action based on data or a decision.
- Communicate: Facilitate two-way information flow between parties.
Most knowledge workers perform all of these tasks throughout the course of their jobs. A content marketer researches a new topic, then creates a blog post.
A sales person researches leads, creates a follow-up email, executes scheduling the demo, then communicates during the live demo meeting.
A CEO analyzes the latest financial forecasts, decides on the company’s Q2 priorities, then communicates them in an all-hands meeting.
These tasks fall on a spectrum: Inputs, Processing, and Outputs.
- Input: Take in information or data.
- Process: Interpret, manipulate, or change data.
- Output: Produce new data, action, or outcomes.
They also tend to differ in how structured they are:
- Structured: Involves a more formulaic sequence of events that can be replicated.
- Unstructured: Involves a more unpredictable or abstract sequence of events.
To analyze (whether text, numbers, images, code, etc.) is a more structured output than to decide (evaluate, judge, select).
Are these categories relative? Of course. Can they vary on a case-by-case basis? Absolutely. To execute can be as straightforward as sending an automated email or as complex as booking a flight through an airline website.
But this simple framework can help you to:
- Divorce the work itself from contexts like department, role, team, or industry.
- Break complex goals and responsibilities into smaller tasks. (This is sometimes called microproductivity, and can help identify subtasks that can be delegated or automated.)
- Consider the role of data in task completion. What information is required to perform a given task?
- Think about the degrees of sophistication and complexity in our work. What do we do that’s routine and logical? What is more abstract and driven by intuition, instinct, or experience?
And most importantly of all, this framework reveals patterns — shared qualities that can help us understand why generative AI might be better suited for certain tasks. (More on that later.)
Who’s the Better Knowledge Worker: AI or Human?
I think people bring unique value to their work: Personal experience. Expertise. Creativity. Empathy.
But I also think some tasks need these human qualities more than others. To use AI wisely, we need to understand which those are. (As Marketing AI Institute first articulated this mission: “More intelligent. More human.”)
It’s not AI doing the magic. It’s AI doing the grunt work, and humans adding the magic.
Next time, I’ll share which tasks are best suited for AI and which are best left with humans.
Meanwhile...
In the meantime, look at your business:
- Can you sort your business operations into these categories?
- Have you found AI more capable of certain tasks?
- Can you replicate these gains across other similar tasks in your company?
I’d love to hear what you uncover! Reach out and let me know 🙂
Question(s) to Consider:
Which tasks take up most of my day? What common qualities do they share? Is there an opportunity for AI to decrease the time or energy they require?
You're Reading a Preview. Want More?
Welcome! This is a preview of our newsletter. Join dozens of marketing and business leaders as we explore frameworks, templates, and advice to navigate the challenges that come with growth. Sign up to receive new insights every other week.
Get a sample policy to jumpstart your company's generative AI guidelines.
How are your employees using AI in the workplace? Do you know?
Download a free generative AI policy to outline AI use and guidelines for your team.
