Speedwalking, Part 2
A few things worth mentioning regarding yesterday's post.
For one, we can become surprisingly skilled at doing the wrong thing.
In February 2025, Canadian race walker Evan Dunfee set a North American 20km record with a time of 1 hour 17 minutes 39 seconds.
For comparison, I once ran a half-marathon (~ 22km) in about 1 hour 20 minutes. An elite walker easily outpaces a regular runner.
In the example of race walking, it's obvious how to go even faster. Just run. In the real world, we can be reluctant to switch approaches precisely because we've become so skilled at the old thing. Not only do we not notice that it's holding us back, but there's also the genuine concern that switching to an ultimately better approach will come with an initial penalty: No matter their athleticism, an Olympic race walker who switches to running wouldn't become an Olympic long-distance runner overnight.
The other follow-up point is that when switching approaches, we can be fooled by unrealistic expectations, thinking that if we just find the right technique, that "one weird trick," everything should feel effortless. This isn't the case. Running certainly isn't easier than fast walking. But it raises the ceiling for how fast we can be based on how much effort we're willing to put in.
And what's that got to do with AI? A few ideas where changing the approach means we stop chasing marginal gains and look for significant impact instead:
You could try to get a computer-use agent to help you fill out all the complicated forms your company requires for this and that. Or you could drastically simplify the workflow in the first place. And then let an AI handle it.
Just... one... more... regex. You're proud of your ETL pipeline and all the manual data transformations and extractions, but with each newly discovered edge case, you have to add another one. Or you can let an LLM handle all of it.
Deploying an AI solution in the first place won't feel effortless. All that data preparation, data cleaning, and evaluation design. But it's worth it because it raises the ceiling and will allow you to go much faster.
Happy running!
