A short podcast by Norbert Korny on the paradox of AI and why if helps us become both more productive and more burned out.
Norbert Korny: AI Makes Us More Productive and More Burned Out
Here is a short video podcast by Norbert Korny I am highlighting as part of my Sunday Spirit blogs.
Transcript:
Now you don’t need to be a maker when you want to build something. An agent can do that for you. And while it’s busy arguing with itself, you can spin up another one.
Hell, you don’t need to stop there. Go ahead and spawn a whole city of them while you’re at it.
But this army of agents produces more within a minute than you can review within a day, and then twice that much on weekends, and I’m sure you’ve heard the bottleneck has moved just enough times by now.
So you removed yourself from building, and now validating becomes your new job. And you might want to validate that you built the right thing because your assistant happens to be half genius, half confused junior.
Still, each unfinished activity you started keeps sitting on top of your head, and it keeps pulling your string even after you stopped working.
Zeigarnik found that you remember open tasks better than the completed ones. Every time you start a new task, you initiate tension.
This tension helps you remember the problem, so it’s easier to jump back in. Finish the task, and the tension is relieved, interrupted, and it persists.
The more things you start, the greater the tension becomes. More unfinished stuff is competing to get back in the game.
And since you can do more using AI, you might feel like you have to. You can keep going during lunch and steal a bit of bedtime later on. And others are doing it too, so it might feel like the right thing to do.
It might feel like you just need to hit one more thing, and then you can get back to your life. This is how the unbearable low creeps in. The people reporting the largest productivity gain from AI happen to be the most likely ones to report burnout.
Aside from the sheer volume of information you need to get through context switching is the mind killer, but switching tasks isn’t necessarily a bad thing.
I find myself like taking a break when switching to a different problem, but there’s still a cost. Part of your attention keeps wandering back to the thing you’ve just left, so even though you moved on, part of your mind hasn’t.
This is what can leave you feeling empty at the end of the day. So we have a couple of options: ignore the task, and it will go away. But it will keep coming back, grabbing at your attention.
Or you can complete it. That gives you closure, and it might look like it’s your only option. But the good news is your brain doesn’t need to have everything done before taking a break.
Sometimes a good next step is enough to tone it down a bit. Make a plan. Seriously, write down when and how you’re gonna pick it back up.
That alone can do wonders for your peace of mind. Now, because AI made the production the easy part–it didn’t magically scale your attention.
There’s no limit on how many agents you can start, but there sure as hell is a limit on how much you can finish. Perhaps start with the most difficult tasks first.
Eat the frog. Eat the biggest frog, and maybe limit the number of agentic frogs you put on your table in the first place.
Keep doing what you love, and don’t let the joy fade.
SKMurphy Take
I agree with Norbert’s perspective and insights, what follows are some refinements and additions.
The bottleneck has moved: Any time you solve one problem that was blocking the neck of the bottle, you promote the problem in line behind it. Progress means you are facing new problems, not the same one over and over.
Zeigarnik Effect: you remember details of uncompleted tasks that are often forgotten after the task is completed. Norbert suggests “making a plan” as one antidote, based on “Consider it done! Plan making can eliminate the cognitive effects of unfulfilled goals,” a 2011 paper by Masicampo and Baumeister in the Journal of Personality and Social Psychology, 101(4), 667–683. They conclude, “Once a detailed plan has been made, one no longer has to think about the goal to execute it. A plan reduces the amount of thought and attention that are typically recruited in service of an unfulfilled goal. Thoughts of an incomplete goal will not interfere with current concerns so long as a plan has been made to see the goal through later on.”
The study was based on a small study of a few dozen college students managing homework assignments, but there is probably merit in “planning the work” if you have the ability to then “work the plan.” The tendency to resume and complete interrupted tasks is called the Ovsiankina effect. A recent retrospective of studies of both effects found strong evidence for Ovsiankina and weak evidence for Zeigarnik.
My personal experience, admittedly anecdotal evidence, is that I am afflicted with both. Writing the same to-do items on a list for days in a row may be necessary, but is not sufficient to complete them. There is undoubtedly satisfaction in finishing a task, even one that has been delayed. A goal you wrote down but left unstarted can be abandoned, but one you have invested effort in is probably better brought to some kind of conclusion. This is especially true of commitments to others.
I am not sure it’s always best to start with the most difficult task first. If it’s a boiled frog, something that is a relatively small task that you have been putting off for no good reason, it’s an excellent idea to take care of that now. But more complex tasks benefit from divide-and-conquer: write a short plan, gather information, run small experiments, and gain a solid understanding of what’s involved.
AI has certainly made us more productive at certain tasks, primarily by providing tireless clerical labor. One way to fight context switching is to find connections and develop a deeper understanding of how different tasks on your to-do list connect to larger strategic goals. If you find yourself working from a “laundry list” of seemingly unrelated tasks, it will be much harder to maintain energy and focus than if you can connect them into a few strategic objectives. If you cannot, then consider pruning unrelated or off-strategy tasks before you start.
I would also make a distinction between trying to solve a “mystery” or address a problem that needs an insight or creative breakthrough where a certain amount of incubation or creative drift is helpful, and situations where you must get down to business and grind it out.
Related Blog Posts
Recurring Problems Have Both Technical and Psychological Roots:“Most recurring problems are a combination of an unsolved technical problem and an unresolved emotional component to that problem.” These are boiled frogs.
Kevin Kelly on Working Smart offers a useful hierarchy of tasks:
- Doing what is required.
- Doing more than is required.
- Trying as many roles as you can in order to discover what you are smart at.
- Making sure you are spending your time on jobs that are effective or that need to be done at all.
- Do only jobs that really need to be done and that you are good at doing.
- Doing work that no one else can do.
Q: How Do I Make Sure I Understand The Customer’s Problem and Present a Vision of a Solution? The reason why most software startups create a product management role (or have one of the founders take responsibility for a coherent product vision) is that taking a survey to gather a laundry list of requirements and treating that as a specification leads to a “Winchester Mystery House” model of construction without architecture.
In PATCA Roundtable on Time Management I offered my 1-2-4-1 model: break a project into four phases, each one goes end to end:
- First 1/8 of time budget: quick overall pass to go end to end and flush out issues and identify key risks.
- Next 1/4 of time budget: a second full pass that focuses primarily on risk areas and missing items.
- Next 1/2 of time: The detailed pass.
- Last 1/8: a polishing pass to help clean up any errors and check for overall conceptual integrity. This method also applies to developing blog posts, presentations, a new product, etc…
See related Chalk Talk on 1-2-4-1 Model
This post Getting Work Done: Leveraging Calendars, Task Lists, and Project Plans also offers a wealth of practical tips
AI related Posts
Image Credit Stepping Stones licensed from https://www.123rf.com/profile_kelifamily