Software engineering is like cycling

There is a famous quote often attributed to Tour De France winner Greg LeMond:

It never gets easier, you just get faster

As you gain resources and become an expert, so does your competition. If we assume effort is a multiplier on potential, any reduction in effort eats away the advantage you gained from previous hard work.

This also means if you're happy to stay at a given level of performance, you do get to "take it easy" to some extent. You'll just be left behind by those who choose not to.

AI tools are like steroids

Years after the Lance Armstrong doping scandal, I heard someone say, in reference to steroids:

It doesn't hurt any less, you just go faster

There are a few interesting parallels. I think anyone who is using generative AI tools to do their work right now should accept that much like using anabolic steroids, they are damaging themselves in exchange for increased performance. In both cases, the impact on the individual widely varies, but I already see former coworkers and friends that are completely unable to work at the level they used to because of their newfound dependence on Claude/ChatGPT etc.

I used to tell people: "If your system falls over when the cache gets wiped, it's not a cache—it's a database." I'm starting to feel the same way about AI assistance. If you can't function without it, it was never assistance.

There are also now "Lance Armstrongs" of programming. Those who are both juicing and putting in the same amount of effort as before. These people are now so productive they often run out of actual work to do and end up shitpost-coding bot-only social media networks and mad max themed job schedulers. I assume these people are also paying some kind of price (on top of the tokens) for this newfound productivity, but it feels far less obvious what.

As for me, I'm trying my best to strike a balance between keeping up with the peloton without hopping on a motorcycle. Personally, I'll be worried when it stops hurting.