Go-to headlines "AI kills developers." This makes sense from a mainstream, dramatic & shareable POV and is wrong. But below it there's a genuine change that everyone hiring engineers needs to realize, since it's altering the trait set for what comprises an excellent one.
AI does not eliminate the necessity for developers. It is altering what a developer will be most valuable for. And if your hiring continues to reward the obsolete thing, you'll be paying fair market for a skill that's increasingly going cheap.
What once was the differentiator
For the longest time a lot of engineering hiring was focused on one thing above everything else — can this person write code quickly, and correctly? Ways of Hiring — Whiteboard problems, coding challenges, take-home tasks. It was mostly just measuring raw production ability.
That was logical when writing valid code was the bottleneck. It was real value; the one who could make more of it, quicker — had true value.
What changed
The AI is pretty good at generating code now. This used to take a human hand an afternoon; now you can generate it in no time at all. The thing that was the differentiating, being able to write code, is becoming the low-cost, high supply part.
The thing you used to compete on becomes a commodity, and the value goes onto some other space. As is the sitting, where it moves bears watching in this instance.
What's getting more valuable
The value is rising in those things that were always more difficult to measure and most of these are human.
- Judgment. Understanding what needs to be built: and, equally importantly, what doesn't. The AI will cheerfully make the wrong thing, beautifully. Everyone else just has to make the decision that it is the right thing!
- Clear thinking. A formal description of the task problem sufficient to enable a human (or machine) to solve it. AI takes vague thinking and magnifies it—vague thinking yields vague results. Clarity becomes a multiplier.
- The question itself: What happens when this goes wrong? before it does. AI gives the impression of certainty and presents ready output. The asset is also the person who interrogates it for an unrecognized flaw [in the story].
- Ownership. One: The desire to be accountable for an outcome, rather than checking a box and passing the baton. When the machine performs the task, taking ownership of the outcome is what's left to a human.
None of these are new. They were always a component of great engineering. What has changed is, they have become more of "the main thing" from being a nice to Have on top of strong coding.
What this means for hiring
The utility of this is that you're looking for something else.
As a result, a candidate who can write perfect code but cannot articulate why something is necessary is less valuable than they were two years ago. Even though their raw coding speed may be totally unremarkable, it is not a scarce resource–coders are now plentiful who are qualified and unqualified alike.
That doesn't mean technical skill ceases to be important. Humans still have to understand the system well enough to evaluate the output of AI. However, "can write code quickly" has lost its title. It is "can judge, think, own".
The junior question
One additional, more difficult issue is worth naming because clever leaders raise it at once. And if AI does entry-level coding, where do junior devs learn their craft?
This is a genuine, tangible problem and halting junior hiring is the typical knee jerk which simply guarantees a shortage of seniors in 2 to 3 years from now. Rather they handle them faster towards the results instead of parking them on routine code for years like we did in previous times. The road to seniority takes a new form, it doesn't vanish.
What this adds up to
The mechnical side of the work is becoming cheap. The human parts: judgment + clarity + ownership → are on the rise. If hiring still optimizes for typing speed, that would be buying the wrong thing.
Has what you actually screen for changed due to the tools your new hires will [actually] use?
No comments:
Post a Comment