Blog post argues engineers may be obligated to use AI despite skill atrophy, just as construction workers lift heavy objects despite joint damage.
Key Takeaways
Using AI to complete tasks reduces learning from those tasks; author grants this but says it may not matter.
The analogy: like carpenters who refuse power tools, engineers who refuse AI may simply be outcompeted out of paid work.
Software engineering’s learn-by-doing feedback loop was a lucky coincidence, not a structural property of the field.
The author compares SWE careers to pro athletes: high earning window, then decline, requiring advance planning.
Unions are unlikely to slow this shift in tech due to high pay, remote work, and global labor supply.
Hacker News Comment Review
Strong pushback on the framing: experienced engineers report that actual coding is 2-5% of the job; the rest is understanding systems and formulating solutions, which AI does not replace.
A key split emerged between engineers who augment reasoning with AI vs. those who replace reasoning with AI; commenters see the latter as genuinely at risk of atrophy.
The abstraction analogy drew criticism: previous abstractions (assembly to C, hand tools to power tools) were deterministic, while LLM output is non-deterministic, making the comparison structurally weak.
Notable Comments
@simonw: flags the title as misleading; the actual argument is narrower: refusing AI may just mean losing paid work, not that engineering careers end.
@hibikir: reports experienced engineers over 40 and 50 using top tooling are outperforming their past selves; AI offloads concentration-heavy tasks where seniors already declined.
@bel8: counters the senior-productivity gain with a direct question: “when a senior can do the job of 6 coworkers, what happens to the coworkers?”
