AI Changed Hiring Faster Than Anyone Expected
Something shifted in hiring over the last two years.
Not gradually. Not in one place.
Everywhere. All at once.
Candidates are using AI to write resumes. To personalise cover letters. To research companies. To prepare for interviews. To practice answers to questions they haven't been asked yet.
On the other side, companies are using AI to screen applications. To score candidates. To conduct first-round interviews. To summarise transcripts. To rank shortlists.
Every layer of the hiring process is being touched.
And yet, if you speak to a hiring manager today - a CTO at a Series B startup, a Head of Engineering at a growing product company, a founder trying to make their fifth engineering hire - the conversation sounds surprisingly familiar.
“We're getting more applications than ever.”
“We're spending more time screening.”
“We're still not confident in the people we're interviewing.”
“We made a hire we regret.”
The process got faster. The confidence didn't.
That gap - between the velocity of change and the quality of outcomes - is what this essay is about.
Everyone Is Solving The Same Problem
The market's response has been significant.
A new category of tools has emerged, each approaching the hiring problem from a different angle.
ATS platforms organise pipelines and manage workflow. AI interview tools conduct first-round conversations at scale. Conversational screening platforms generate structured transcripts for recruiters. Candidate intelligence platforms aggregate what already exists about a person - LinkedIn profiles, GitHub repositories, digital presence - into a single evidence file. Skill assessments and work simulations put candidates into controlled environments to measure performance.
Each is a genuine attempt to solve a real problem.
Each is responding to the same underlying reality: traditional hiring signals are weakening.
A resume written with AI assistance looks different from one written without. An interview answer prepared with AI coaching sounds different from one that isn't. The signals that hiring teams have relied on for decades are becoming harder to read.
So the industry built more tools to generate more signals.
The industry's response was logical. The hiring problem simply evolved faster than the solutions.
The Company Problem
For most of hiring's history, the challenge was straightforward.
Not enough information. Not enough visibility. Not enough candidates.
The resume was designed for a world where information was scarce. It compressed a person's professional life into a page because that was the most efficient way to transfer relevant information from candidate to company.
Interviews existed to fill the gaps the resume couldn't cover.
The system was imperfect, but the logic was sound: gather as much information as possible, then decide.
Today, that logic has inverted.
A company posting a senior engineering role receives hundreds of applications within days. Each resume looks more polished than the last. Each cover letter is articulate and relevant. Each candidate profile is optimised for discoverability.
The volume of information has increased dramatically.
The confidence in that information has not.
Hiring managers are not struggling to find candidates anymore.
They are struggling to trust what they see.
A resume that claims five years of backend engineering experience - is that five years of deep ownership, or five years of peripheral involvement? A candidate who interviews well - is that genuine capability, or practiced performance? An application that perfectly matches the job description - is that genuine fit, or keyword optimisation?
The problem isn't volume.
The problem is trust.
More information without better signal doesn't produce better decisions.
It produces more uncertainty dressed as confidence.
The Candidate Problem
Most articles about hiring stop at the company.
That's the wrong place to stop.
The hiring problem is bilateral. Both sides of the table are paying the price.
A resume captures outcomes, titles, and technologies. It struggles to capture judgment, learning, trade-offs, failures, and growth - the experiences that actually shape how someone thinks and works.
This affects experienced engineers, whose depth of reasoning and ownership is flattened into bullet points. It affects candidates earlier in their careers even more acutely - people who have developed genuine capability through constraint and difficult environments, but who get filtered out not because they lack ability, but because the system was never designed to see what they actually have.
Companies are making decisions with incomplete signal.
Candidates are being evaluated on an incomplete representation of who they are.
What a better representation might look like is a question worth asking - but one for another time.
The Shift Nobody Talks About
Here is what changed - and what most hiring commentary misses.
Ten years ago, the hiring industry asked:
How do we get more information about this candidate?
That question has largely been answered. The information exists. LinkedIn profiles, GitHub repositories, public footprints, AI screening transcripts, digital presence - all of it is now accessible, aggregatable, and deliverable in seconds.
Today, the industry is asking a different question:
I have more information than ever. Which of it should I trust?
This is not a small shift.
It is the difference between an information problem and a signal problem.
Information is data that exists.
Signal is information that predicts something.
Learning is understanding which signals actually mattered.
Every improvement in hiring over the last decade has made information easier to collect. The next decade may be defined by how effectively we learn from it.
The hiring industry spent a decade solving the information problem.
It is now discovering that more information does not automatically produce more signal.
It can produce the opposite.
Every hiring process generates signal.
Very few generate learning.
That distinction is where the next chapter of hiring begins.
The Missing Outcome Loop
Every hiring system in use today has the same architectural gap.
It ends at the hire.
The resume is reviewed. The interviews are conducted. The assessment is completed. The scorecard is filled. The offer is made. The candidate joins.
And then the hiring system moves on to the next role.
What happens after - whether the hire succeeded or struggled, whether they stayed or left, whether the signals that led to the decision actually predicted performance - is never fed back into the system.
This is not a minor gap.
It is the central gap.
Without outcome data, every hiring cycle starts from zero.
The hiring manager who made a great decision last year has no structured way to understand which signals led to that decision. The hiring manager who made a poor decision has no structured way to understand where the evaluation went wrong.
The same frameworks are applied to the next hire, with no learning from the previous one.
Consider what this means at scale.
A company that makes twenty engineering hires a year has, over five years, made a hundred hiring decisions. Some succeeded. Some didn't. Some signals correlated with success. Others didn't.
That information exists - in performance reviews, in manager feedback, in attrition data, in promotion records.
But it is almost never connected back to the hiring signals that preceded it.
The questions nobody is tracking:
- Did this person succeed in the role?
- Did they stay beyond twelve months?
- Did they grow into greater responsibility?
- Which signals in their evaluation predicted that outcome?
- Which signals were present but irrelevant?
- Which signals were absent but turned out not to matter?
Without answers to these questions, hiring remains a high-stakes activity conducted largely on intuition, pattern recognition, and hope.
AI has accelerated evaluation.
It has not yet accelerated learning.
What Comes Next
The hiring industry has built remarkable tools for generating information.
It has built increasingly sophisticated tools for structuring and presenting that information as signal.
The next step - the one that has barely begun - is building systems that learn from what that signal actually predicted.
Not systems that generate more reports.
Systems that ask: did we get this right? And if not, why?
The properties of such a system are not difficult to imagine - something that evolves with each hiring cycle, that connects evidence to outcomes, that builds understanding over time rather than resetting with every new role.
What that looks like in practice is still being worked out.
But the direction is clear.
Hiring has moved from an information problem to a signal problem.
The next move is from a signal problem to a learning problem.
The Aikiyam View
At Aikiyam, this is the question we're exploring.
Today, we conduct structured deep-dive conversations with engineering candidates - focused on real projects, real decisions, real trade-offs, and real ownership. We capture what we know, what we don't know, and where additional validation is needed. We deliver that to hiring teams before interviews begin.
And when a hire is made, we track what happens at 30, 60, and 90 days - not to measure the candidate, but to understand which signals predicted success and which didn't.
We're exploring one part of a much larger question.
Signal creates reports.
Signal connected to outcomes creates intelligence.
AI changed how we collect hiring signals.
The bigger opportunity is learning which signals actually predict success.
Continue the Conversation
These essays document the thinking behind Aikiyam as we build it in public. If you want to shape the future of hiring with us, we'd love to hear from you.