Tokenmaxxing and the Illusion of Productivity
Doing more is not always better
When I think of some of the most asinine trends recently, tokenmaxxing is way way up there, but psychologically I can see how we got here.
As employers continue to create fear driven cultures around AI layoffs, workers will trend towards what creates the perception of “value creation” and safety. At times, this will be done regardless of the “actual value” of the work being done.
You don’t have to look far to spot this behavior in action.
So, what are we really doing here? What's being optimized for? Who are we trying to please (fool)?
Look, I do not actually know where to put myself on the AI spectrum at the moment. I am someone who has actively driven to put AI strategies and structures into place from zero in multiple organizations.
I have half a dozen solo projects of varying levels of seriousness to learn new skills, understand emergent patterns, prove concepts and solve problems, small and large.
I leverage Claude and Perplexity in day to day life for things well beyond research or cited web searches and towards more of an embedded operating system to pieces of life.
That said, things I have always valued is using the right tool for the right purpose and intent and being efficient.
The sheer idea of “maxxing” is antithetical to how I think in general, so this always rubbed me the wrong way. That said, I had to dig deeper to ensure that it wasn’t some strange FOMO driving the sentiment.
The longer I sit with it, the more I’d like to think that I understand it.
People are grasping at straws right now. They want to keep their jobs. They want to appear “innovative.” We are trained to chase rewards. All of us have watched people actively not solve the real problem at hand because of perverse incentives at play, and that’s exactly what I see happening in many businesses right now.
To create a healthy culture of AI adoption, I truly believe that it’s a very human thing you must tap into. Find real problems, teach people skills, scaffold them so that it doesn’t feel daunting, scary or out of reach. And, most critically, make it clear what the objective at hand is.
I have watched many people and companies completely lose sight of any actual goal to solve a problem or help a customer and simply obsess about tool adoption and usage.
My favorite analogy thus far is that we’re watching an Uber Eats completion for miles driven with no regard for meals delivered. You solve no customer pain, but you sure did do a lot of stuff.
Keep it real simple. Solve a problem, care about people.
AI should be used to let people do the job they were hired for at their maximal level of impact. How much time spent on toil can you reduce for an engineer, how much time saved on data synthesis, problem diagnosis and triage etc. to get to actually solving a problem.
Move what you measure, adjust what you reward and the behaviors will follow.
If you want a counterpoint to this perspective which is arguing for aggressive but pragmatic adoption of these new tools which will and are fundamentally reshaping the way we work, look no further than this.
It’s an understandable position, if not a bit extreme in approach. I think this is akin to saying “We’ve never had cars before. Here’s a car. Now, everyone is going to drive as much as possible all the time, no matter where, for what, how fast, safely or in what direction. Just go. Just keep going until the wheels fall off.”