If you’ve been on the internet at all over the past 10 years, you’re probably familiar with the “one weird trick” ads.
These are apparently called “Chumbox ads” - which, according to Wikipedia:
“promise easy solutions to complex medical or financial problems, use sensationalist language like ‘one weird trick’ and ‘Prepare to be shocked!’ or allege that the government or powerful industries ‘don’t want you to know’ about their products.”
It’s lowest common denominator stuff. Get rich quick, lose weight fast, be successful with this one weird trick. A lot of AI stuff has swerved into one weird trick territory:
A few days ago the Harvard Business Review ran an article on how AI-generated “Workslop” is destroying productivity. Essentially, workers are using AI to be so ‘productive’ that it’s overwhelming their coworkers or supervisors with the work of figuring out whether or not what they’ve produced is actually worthwhile. Workslop is, simply put, outsourcing critical thinking to AI and having it churn out pages upon pages of analysis and conclusions. From the article:
Here’s how this happens. As AI tools become more accessible, workers are increasingly able to quickly produce polished output: well-formatted slides, long, structured reports, seemingly articulate summaries of academic papers by non-experts, and usable code. But while some employees are using this ability to polish good work, others use it to create content that is actually unhelpful, incomplete, or missing crucial context about the project at hand. The insidious effect of workslop is that it shifts the burden of the work downstream, requiring the receiver to interpret, correct, or redo the work. In other words, it transfers the effort from creator to receiver.
An example: imagine asking a new associate in a law firm for a memo on a client’s liability risks given a particular set of facts. 15 minutes later the associate walks back into your office with a polished-looking 12-page memo and legal analysis, complete with end-notes. That new associate is certainly going for high productivity, but what you have in your hands is also probably some high-grade workslop.
The difference between value and productivity:
There is a big difference between being productive and creating value. If you optimize for sheer output - measuring and rewarding the number of things done in a set amount of time - you can end up producing a lot of things that aren’t actually valuable. The practice of law isn’t immune, especially when incentives on value to the firm and value to the client are out of whack1. In some other industries productivity and value are more closely correlated - like in a factory where the output is, say, a widget that’s in high demand. As long as you can sell those widgets, the faster you can crank ‘em out, the more you can sell, the more valuable the widget production line becomes.
But back to our new associate. We need to consider what the point of creating these memos is in the first place. Are they valuable in and of themselves? I think I’m quoting or maybe paraphrasing Ethan Mollick in this, but think about a high school English teacher giving their students an assignment to write a 5-paragraph essay. Is the teacher giving them that work because the essays are valuable in and of themselves? Is the high school teacher rewarded in some way for the sheer number of 5-paragraph essays the students produce? Or does the value lie in making the students go through the process of writing?
Easy or hard:
Another example: I set up a scheduled task in both Gemini and ChatGPT that runs every Friday morning, where they search for and read news articles that have to do with law + AI over the past week. They create an overview of them, identify themes, and give me a nicely formatted report. It’s kind of magical, but has some distinct flaws. Here’s a sample excerpt from Gemini:
The stories from the past week reveal a legal industry at a crucial inflection point, where the theoretical promise of AI is rapidly becoming a practical reality. The most significant revelation is that AI adoption is not confined to well-funded “BigLaw” firms; it’s being embraced by organizations tackling fundamental societal problems, like the justice gap. This suggests that the transformation AI brings will be far-reaching, potentially democratizing legal services and reshaping the profession from the ground up.
This is eye-rollingly grandiose. Frankly I don’t want to read these kind of sweeping conclusory statements from anyone - much less Gemini, and will have to go re-work the prompting for it, but I may not be able to keep it from doing this. It’s productive, sure, but not actually valuable in the way I want. In a sense this is workslop of my own making.
To grossly oversimplify things: with AI increasing productivity is now easy, but increasing value is still hard. Last time I wrote about the difference between “good” and “quality” - and maybe a better way of distilling that is by distinguishing between sheer quantity (productivity) and the quality (has value):
I think downplaying this by saying “AI simply isn’t good enough” misses the broader point: there’s always a difference between what’s good and what’s quality. AI is astonishingly good at making things, even podcasts, but that doesn’t mean it’s making things of quality.
Creating value is the hard part
The problem with workslop is not with AI itself - it’s with the workers and managers who mistake productivity with value. It’s almost like clicking a “one weird trick” ad - “Employees are suddenly 10 times more productive with this one weird trick.”
Let’s go back to our new associate example. Are the liability analyses created by brand new associates actually valuable, in and of themselves? There’s value in the billable hours, sure - but often those get written down for clients. There’s some value in having a fresh take on liability given a set of facts, but the client probably isn’t keen on having their decision to settle or fight be based on the take of a brand new associate. So why do them? Maybe the real value lies in the new associate getting the reps on the actual practice of law, in the hopes that one day that new associate becomes a partner and is the one clients trust with questions about liability, and therefore gets more and more matters for the firm. Doing these analyses may not be valuable in the immediate term, but just like exercise or dieting, the payoff is in the reps.
It should not be Turbo Time
Reading about workslop makes me feel like people are taking the famous quote that people don’t actually want to buy a quarter-inch drill, instead they want a quarter-inch hole, and misapplying it in such a bad way that we’re running through the walls like it’s Turbo Time.
Maybe the hard part to creating value is figuring out what has actual value. There really is no “one weird trick.” Instead of answering the question of why people want to buy drills with “holes in walls,” we should maybe ask them why they want the holes in the walls in the first place…
The billable hour is a prime example of this. Just saying.