Something significant is happening. Not eventually – now. If you work in talent acquisition, you have already felt it: AI tools that draft job ads in seconds, parse thousands of CVs before your morning coffee, and schedule interviews without a single human hand touching a keyboard.
In a viral essay published earlier this year by Matt Shumer, AI entrepreneur and CEO of HyperWrite, reframed the conversation in terms that are hard to ignore. In “Something Big Is Happening,” Shumer argues that the newest generation of AI models are not merely better tools – they are functioning as autonomous workers. He writes about AI completing his own technical tasks to a standard meeting or exceeding his own, leading him to a disquieting conclusion: the disruption of cognitive work is not a future event. It has already begun.
Shumer’s framing matters for recruitment professionals for a simple reason: if white-collar cognitive work is genuinely in transition, then the industry that finds, evaluates, and places people into that work sits at the very epicentre of the disruption. The question isn’t whether AI will change recruitment. It’s whether we will shape that change – or be shaped by it.
The numbers are hard to argue with
Adoption is no longer an edge-case story. A 2025 McKinsey report found that 78% of organisations are using AI in at least one business function, and recruitment is among the fastest-moving areas. Roughly 70% of employers currently rely on AI to automatically screen out applicants in early hiring stages (HR Grapevine). The market has moved from experimentation to normalisation in a few years.
What’s driving uptake? Volume, mostly. A mid-sized company posting a visible role today can receive thousands of applications within days. No human team can process that meaningfully at scale. AI offers a way to do initial triage – summarising CVs, identifying keyword matches, flagging candidates who meet baseline criteria – that would otherwise require an army of coordinators.
The efficiency case, when AI is used well, is genuinely compelling. A Stanford-led field experiment, published by the World Economic Forum in 2025, found that candidates who went through an AI-assisted interview pipeline succeeded in subsequent human-led interviews at a rate of 53%, compared to just 29% in the traditional CV-screening track.
The distinction matters, though: AI worked here as a pre-filter that surfaced stronger candidates for human judgment, not as a replacement for it.
The Shumer thesis: When automation meets autonomy
What sets Shumer’s argument apart from the usual tech-optimism cycle is the specificity of his claim. He isn’t predicting a future state – he is describing a present one, observed from inside a company building the very tools he’s describing. His point is that AI has crossed a threshold: from assistive to autonomous, at least in domains that are screen-based, cognitive, and language-driven.
“If your job happens on a screen – reading, writing, analysing, deciding – then AI is coming for a significant part of it. The timeline isn’t ‘someday.’ It has already started.” – Matt Shumer
His words carry weight precisely because they are not abstract. He describes AI completing his own technical work to a high standard, leading him to write: “I am no longer needed for the actual technical work of my job.”
For recruiters, this should prompt a genuine reckoning. If autonomous AI can handle the reading, writing, analysing, and communicating that defines most desk-based work, then which parts of the recruitment function remain distinctively human? And more pressingly: what should we be doing to ensure those parts are not quietly engineered away in the name of efficiency?
Where AI genuinely adds value
Let’s be precise about what AI does well in hiring – because getting this right is the foundation of a sensible hybrid strategy.
- High-volume screening: Parsing thousands of applications against a job brief, flagging relevant experience, and sorting candidates into priority tiers is exactly the kind of pattern-recognition task at which current AI excels.
- Scheduling automation: Eliminating the back-and-forth of interview coordination saves recruiter hours and reduces candidate drop-off from friction.
- Job description and ad drafting: the right AI tools can produce a solid first draft of a job description and job ad in minutes, freeing recruiters to focus on refinement and specificity rather than blank-page anxiety.
- Bias auditing (when applied correctly): Structured AI evaluation, applied consistently, can reduce certain forms of unconscious human bias – though, critically, only if the training data and weighting criteria are themselves audited.
- Candidate communications: Automated updates at key stages keep candidates informed and reduce the anxiety of silence, which is one of the most common complaints in candidate experience surveys.
These are legitimate efficiencies. The problem arises not from using AI for these tasks, but from extending it into territory it is not well suited for – and then failing to say so.
The risks nobody puts in the brochure
AI in recruitment has a set of well-documented failure modes that deserve more honest attention than they typically receive in vendor materials.
Algorithmic bias is the most serious. AI hiring tools learn from historical data. If past decisions favoured certain demographics – certain educational backgrounds, institutions, or even names – the model will likely encode and amplify those patterns. Amazon’s notorious case is instructive: the company abandoned an AI screening tool after discovering it had systematically penalised CVs that included the word “women’s.” A 2022 MIT Sloan study found that AI-based CV filters reduced interview rates for women in tech roles by around 15%. These are not hypothetical risks.
The loss-of-nuance problem is subtler but equally real. Algorithms struggle with context. A candidate who took two years away from paid employment to care for a seriously ill parent may be flagged as having an “employment gap” without any mechanism for the system to register what that gap represents. A non-linear career path – someone who moved from teaching into operations, or from journalism into product management – may score poorly against a keyword-matching rubric designed for conventional progression. The unconventional candidates are often, in practice, the most interesting ones.
Then there is the transparency gap. When a candidate is screened out by an AI system, they typically receive no explanation. Under emerging regulation this is becoming a legal issue as well as an ethical one. The EU AI Act classifies AI-based recruitment tools as high-risk systems, requiring documented human oversight. The US Equal Employment Opportunity Commission has issued guidance warning that reliance on biased algorithms can expose organisations to disparate-impact claims. The regulatory direction of travel is clear: less automation, more accountability.
What the data tells us about what people actually want
Here is the tension at the heart of the current moment: AI adoption is accelerating, but so is discomfort with it. At a recent SocialTalent Live event, over 80% of participants said they wanted more balance between human and AI input in their hiring process. Nearly 70% of candidates in separate research stated that they are uncomfortable with AI making hiring decisions, and the majority expressed a clear preference for companies to be transparent about when and how AI is involved.
People want to feel seen. That phrase – unremarkable in its simplicity – contains a principle that no algorithm has yet managed to operationalise. The moments that leave a lasting impression in a hiring process are almost always human ones: the recruiter who took time to understand a candidate’s situation, the hiring manager who gave real feedback after a rejection, the coordinator who picked up the phone when something went wrong.
A recruiter who spoke to a candidate nervous about maternity leave was able to address concerns about start dates, flexibility, and travel arrangements – concerns that would never have surfaced in an automated screen. The candidate felt in control. They performed well. They got the job. That outcome was not achievable by an algorithm. It required someone to slow down, listen, and respond to what was actually in the room.
The irony organisations need to sit with
There is a particular irony in play right now that deserves to be named directly. Many of the same organisations investing heavily in AI-driven hiring automation are simultaneously running culture initiatives, belonging programmes, and engagement surveys premised on the idea that people are their most important asset.
The first interaction a candidate has with your company is your hiring process. If that process is impersonal, opaque, and automated from start to finish, what message does it send about how you treat people once they are inside? The candidate experience is not peripheral to your employer brand. It is your employer brand, made real.
Using AI to process people efficiently is not inherently incompatible with treating them with respect. But it requires deliberate choices – about where human contact is inserted, how transparency is maintained, and what happens when someone has a question that no chatbot is equipped to answer.
A practical framework for human-AI balance
Getting this right is less about philosophy and more about process design. Here is a framework that talent leaders are finding useful:
- Use AI for volume, humans for judgment: Let AI handle the tasks where volume and consistency matter – initial screening, scheduling, first-pass communications. Preserve human decision-making for the moments that carry weight: which candidates move forward, how interviews are conducted, and how offers are framed.
- Audit regularly, not once: Bias in AI systems is not a one-time problem that gets fixed at deployment. It evolves as hiring patterns evolve. Schedule quarterly reviews of screening outcomes by demographic cohort, and be willing to adjust or pause tools that are producing unacceptable patterns.
- Tell candidates what you’re doing: Transparency is not just an ethical preference – it is increasingly a legal requirement. Be explicit about where AI is used in your process. Give candidates the option to request human review at key stages. This is not weakness; it is trust-building.
- Inject human contact at milestone moments: The first interview invitation, the final decision, the offer, the rejection – these are the moments that define how a candidate feels about your organisation. Automation can handle the notifications. A human should handle the meaning.
- Invest in recruiter upskilling, not just AI tooling: The promise of AI in hiring is that it frees recruiters to be more human – to spend more time on meaningful conversations, relationship building, and candidate guidance, and less on administrative processing. That promise is only realised if organisations actively invest in the human side of the equation, not just the technology.
The questions worth asking before your next tool purchase
Before implementing any AI recruitment tool, there is a short checklist worth working through honestly:
- Does this make candidates feel more or less valued?
- Would we want to go through this process ourselves?
- Are we solving for our convenience or for genuinely better hiring outcomes?
- What are we losing when an algorithm replaces a human judgment call?
- Is this the first impression we want to make on someone we are hoping to hire?
If the honest answers to these questions are uncomfortable, that discomfort is data. Act on it.
The larger stakes
Shumer’s essay ends with a call to take the transition seriously – not with panic, but with clear eyes. He is right. The pace of AI development in cognitive work is not slowing. The organisations that will navigate this well are not the ones that automate most aggressively, nor the ones that resist most stubbornly. They are the ones that think carefully about what humans are irreplaceably good at – and design systems that protect and amplify that.
In recruitment, that means recognising that the goal is not to process candidates. It is to find the right people and give them a reason to choose you. That is a fundamentally human project, whatever tools you bring to it.
The algorithm can screen. It cannot see. And seeing – really seeing what someone is capable of, what they bring beyond their CV, what they might become – is still the job.
