Leadership1 publisher2 min readPublished
A hiring platform's CEO wants first-round engineer screening moved inside a simulated codebase
The Gartner numbers behind the argument count how many candidates use generative AI. The one-in-four-fake-profiles figure is a forecast for 2028, and the wasted senior-engineer hours arrive without a count.
The Board Room · Leadership desk
What happened
- Gartner's researchers say around 39% of candidates now use generative AI as part of their job application process, according to a Forbes Tech Council article on technical screening.
- The same researchers predict that by 2028, one in four candidate profiles worldwide will be fake.
- The argument comes from Adithyan RK, chief executive of the company Hyring, writing as a Forbes Tech Council contributor.
- His recommended replacement for document screening is a containerized workspace copying an enterprise stack, holding a broken REST API, an unoptimized database index and a failing integration test suite.
- The piece also cites unnamed industry research saying most companies have put AI into recruiting and that 93% of recruiters planned to boost their use of AI tools this year.
Compiled by The Board RoomSomething wrong?How this is made
Why it matters
- contradiction The article warns that rejecting candidates for using AI to write a resume may be the wrong strategy, while also saying the resume has lost all predictive value; a screening rule cannot follow both, and which one a team picks decides whether it filters resumes or ignores them.
- decision A leader setting screening policy this quarter is acting on an adoption figure, because the fake-profile number describes 2028 and not the funnel in front of them.
- cost Swapping document screening for simulated codebases and telemetry proctoring moves spend from senior engineer hours onto a platform contract, and the case for that trade rests on an unmeasured "dozens of hours".
- exposure Proctoring that analyzes behavioral telemetry brings candidate monitoring inside the hiring process, and the article covers that method in a single line rather than its consent or compliance handling.
The two Gartner percentages in the article's central sentence do not fit together as printed. Around 39% of candidates use generative AI somewhere in the job application process [2]. The same sentence puts resume drafting at 54% of candidates [3]. Drafting sits inside the application process, so the part is 15 points larger than the whole [12]. One reading reconciles them: 54% may be a share of the candidates who use AI at all. As written, both figures are shares of candidates.
The diagnosis and the remedy come from the same author. Adithyan RK, chief executive of Hyring, wrote the piece for the Forbes Tech Council [1]. His remedies are sandboxed engineering environments, which spin up a containerized workspace replicating an enterprise infrastructure stack, and proctoring suites that analyze behavioral telemetry [10] [11]. The article does not say what Hyring sells.
Hours are the claim an engineering leader would act on. Engineering managers are "wasting dozens of hours interviewing candidates who look like principal architects on paper but struggle with basic technical execution in practice," RK wrote [6]. That is an estimate without a measured sample or interval behind it.
Two of his sentences point in opposite directions. One says that once a resume can be produced in under 30 seconds, "the document loses all predictive value" [5]. Another says "turning down candidates just because they employ AI for resume writing may not be the best strategy" [8]. A screener has to act on one or the other.
The two headline numbers cover different periods. 39% describes behaviour now [2]. The one-in-four-fake-profiles figure is a prediction for 2028 [4], roughly two years after the article ran [14].
One recommendation is a change of policy. RK tells executives to decouple pedigree from performance, on the argument that a prestigious past employer or an Ivy League degree is no longer a guaranteed proxy for immediate technical capability [15].
RK's remaining recommendations divide by cost. His assessment examples are a flawed pull request to review, an inefficient database query to optimize, and a bottleneck the team recently solved to architect around [9]. A senior engineer can write those in an afternoon and reuse them. The containerized environment and the telemetry proctoring are procurement. On the employer side of the same funnel, the piece credits "industry research" for the finding that 93% of recruiters planned to boost their use of AI tools this year [7].
What to watch
- Whether Gartner publishes the base for the 54% figure, which would settle whether it counts all candidates or only AI users.
- A measured before-and-after on senior engineer interview hours from a team that moved first-round screening to work samples.
- Whether proctoring that analyzes candidate behavioral telemetry draws attention from an employment or privacy regulator.