AI is in a bit of a weird place right now.
On the one hand, advancements keep coming — every HRIS and ATS is scrambling to introduce new AI tools, search algorithms are changing, and AI advocates keep making bigger promises.
On the other hand, those big promises haven’t been kept. Every company has a failed AI pilot or is only using it at a small scale, with none of the life-changing innovations originally promised.
Or, as our moderator Hamraj Gulamali put it: “The consumer uptake of ChatGPT has been huge… and yet at the business level we haven't quite seen that transfer over. People are using it more in their day-to-day lives than in their professional lives. Speed is not easy when going through technological revolution.”
So what’s really going on?
On Wednesday, 3rd December, we brought together a panel of experts to chat about the future of AI for talent teams:
- Reality vs optimism: How fast will AI for talent evolve?
- Why agentic AI is already changing the job of recruiters
- What productivity looks like in the age of AI
The panel:
- Hung Lee, former recruiter and founder of Recruiting Brainfood
- Kasia Hayward, Head of Culture and Innovation, Atlas
- Daniel Harrison, AI specialist at Amazon Web Services (AWS)
- Hamraj Gulamali, moderator and Zinc’s Head of Legal and Compliance
1. Reality vs optimism: How fast will AI for talent evolve?
Daniel, whose work is embedded with all things AI, told us he uses around 15 AI agents every day. But Amazon is uniquely positioned to experiment and push the boundaries of what’s possible.
So we asked Hung: Can a 500-person company replicate this for all their employees? “Amazon is a unique business, so it’s almost incomparable to everywhere else,” he pointed out. “Is a standard company that isn’t doing software going to be able to deploy agentic AI in the same way? Obviously not.”
He went on to point out that, while these companies might be experimenting, they’re waiting for AI to be brought to them by a vendor, not create it themselves.
As a counterpoint, Daniel pointed out that pre-built applications are rolling out more agentic tools that let users create agents that help them execute tasks using natural language, and the capability is coming within the next year.
Where they might have needed deep technical skill, now anyone can build in tools like Amazon Quick Suite, Google Gemini, and Microsoft Copilot. And with more democratisation comes more opportunities for everyone to harness the power of AI.
“Amazon is a preview, rather than a statistical outlier,” he said.
The people implications: Speed and trust
In the next weeks, months, or years, this tech might be readily available — so what does it mean for your people team, or for building a high-performing workforce?
At Atlas, Kasia’s team partnered with AWS as they were developing a platform for people leaders that leaned on human identities and anthropological research. “On the one hand, we want to be really innovative,” said Kasia. “We want to use the AWS bells and whistles.”
But when they asked clients, they got mixed reactions. “They said, ‘I really like this, but can we switch the AI feature off?’” Kasia recalled.
Using AI is going to affect business and data, and it’s important to think about who is going to benefit and what the outcomes are to avoid getting overexcited or overly worried.
When AI arrives too quickly, trust becomes an important factor in your relationships with clients and employees, and can have serious implications for candidate experience.
What sectors are going to keep up, and which ones are going to fall behind?
Not every industry moves at the same pace. When our panel turned to think about which ones would struggle to adapt to an agentic-first world, Daniel pointed out that it would probably be more traditional industries.
Spaces like manufacturing and the public sector tend to move slowly. “Part of the reason is because they don’t necessarily have the right culture in place through which to continuously drive that innovation,” he added.
At Amazon, every employee can build and deploy into production if it’s worth it. In other industries, though, ideas and innovation have to work up the hierarchy and are often outdated by the time they get approved.
“It’s not a technology challenge. It’s 100% a culture challenge,” he said.
The gap between agency and in-house adoption
Often, candidates are adopting AI faster than talent teams can keep up. But is there a gap between in-house teams and recruitment agencies?
“Agencies are probably ahead, mainly because of a higher risk tolerance and less concern about compliance,” said Hung. “Some small boutique agencies are doing things 6 months ahead of what a TA team can do, because TA has more responsibilities.”
An in-house talent team has to think about ethics, candidate objections, and all the other barriers that come from working in a wider business.
But even if agencies are ahead, it’s not by much — Hung reckoned that one 12-week sprint could get in-house teams on par with most agencies.
2. Why agentic AI is already changing the job of recruiters
If you’re being flooded with AI-generated applications, you’re not alone. Just as TA teams are learning to use AI agents to book meetings, sort through candidates, and take over other manual tasks, candidates are also using AI to apply to roles that may or may not be relevant for them.
So who’s going to win?
Hamraj asked the panel: Are we just moving to a world where AI talks to itself, and humans are cut out of the process?
Daniel challenged the question. As organisations get access to new technology, everyone has the right to use it. The same goes for candidates — they’re just trying to use similar tools.
“It’s not inherently bad,” he said. What is a risk is when losing human oversight and values.
“The question becomes: how do we insert human values into these agents?” asked Daniel. Whether it’s ethics or security, teams need to take into account:
- Emotional intelligence
- Compliance standards
- What tools are available, and to whom
- Identity access management
Agent to agent communication doesn’t mean that there’s no role for TA. It means that TA teams have to rethink the hiring process in response.
Balancing candidate consent and risks
Kasia pointed out that candidates are already susceptible to fraud during the hiring process, and we need to re-think the risk of how candidate data is processed. Whether it’s getting consent from candidates or making sure the hiring process is still fair, there’s always nuance that human recruiters need to pick up on.
“It’s really important that you design your talent strategy with AI and innovation, but also ring-fence it,” said Kasia.
The problem with AI fraud
There are a lot of headlines about fake candidates and the fraud risk they pose for companies. But is it actually a problem, or just a marketing strategy?
Hung pointed out that the risk isn’t equal across all companies, but varied based on industry and size. “If you say you’ve never seen fake candidates — you’re probably not a tech company, not remote, not a target. Fraud risk surface area depends entirely on who you are and who you recruit,” he said.
Hiring for in-person roles doesn’t carry the same type of risk. Tools that can detect fraudulent candidates aren’t perfect for everyone, but some companies definitely need it.
Updating DEI and re-thinking bias
Just like people, AI models have biases. But they can also be a powerful force for reaching underrepresented groups and using the right language.
Defining what you want, as a business, out of DEI efforts is key, said Kasia. “At the end of the day, you think in quite abstract terms and it’s difficult for functions to quantify,” said Kasia.
As an example, Kasia pointed to LinkedIn. “Senior women are reporting much lower impressions and engagement on LinkedIn,” she said. “Some are changing their gender in posts to meet the new algorithm criteria. That’s the challenge — if AI is trained on certain data and goals, we may get something we don’t want.”
The data AI models are trained on shapes the output, and TA teams have a responsibility to make sure that the data that goes in is as unbiased as possible.
3. What productivity looks like in the age of AI
Hung made the observations that AI is so new, even predicting three months out is difficult. Some people are optimistic, some people are pessimistic, and some people swing wildly between the two positions.
“There are a lot of smart people who don’t know what’s going on,” said Hung, “and that gives me hope. The future isn’t fixed, which means we can influence it.”
But if organisations don’t do anything about AI, it’ll start impacting businesses. The biggest cost to any business is payroll, and funding is going towards companies who can use workforce automation tools.
Hung already sees it having an impact on the bottom level of talent. “I think the junior sourcer has already gone. I can’t remember seeing one advertised anywhere.”
The positives?
- Everyone’s in the same boat
- AI is still ambiguous enough that teams can influence it
- The way to influence it is to think differently about the politics of productivity
How do TA teams make sure AI adoption is fair?
If the promises of AI are fulfilled, they’re going to result in one thing: more time.
And if you manage to find more time through AI automation, the most responsible way to respond is to use that time somewhere else. So how do you use it properly to boost employee productivity?
“In recruiting, we care about candidate experience, DEI, and all these wonderful things,” said Hung, “but they’re the first to go when the efficiency conversation comes.”
Instead of removing human-first programs in the name of efficiency, AI can help make a case for reinvesting the time saved back into programs that recruiters know makes candidates and employees happier.
Building relationships with candidates is at the heart of talent acquisition. Instead of outsourcing that to AI, use AI for more opportunities to build that connection.
When a member of the audience asked Daniel what he did with his time after building 15 different AI agents, he didn’t hesitate.
“Now I’m reallocating my time towards interpersonal work. Executive briefings, whiteboarding, problem-solving with customers,” he said. “For interviews, what used to take 90 minutes now takes a 3-minute voice note. The agent drafts the feedback. I review and submit. That frees me to focus on face-to-face interactions.”
Redefining the role of HR teams
As anyone in HR or talent can tell you, teams are responsible for a lot: legal, compliance, talent, DEI, wellbeing… the list goes on.
“The people function is bursting at the edges,” said Kasia. “AI will reduce admin — but that forces you to redefine what HR is, not just pile more on because tech makes us ‘more productive’.”
Making sure that teams have time to be strategic and make a measurable impact on the business is key to defining what talent looks like in the future.
It doesn’t mean piling on more tasks in the name of efficiency. “In the future, teams will have to try and think deeply about what they look like and how they’re composed,” said Kasia.
AI isn’t good for everything
To wrap up, Hamraj asked Daniel what he thought AI agents would still be bad at in a year.
Daniel pointed out that AI is great at automating and removing repetitive tasks, but it’s not great with subtlety.
“AI can understand words, but it can’t read nuance — the facial expression, the hesitation, the emotion. As long as it can’t do that, there will be elements of HR and Talent that remain fundamentally human,” he said.
AI will never replace a recruiter’s ability to navigate a difficult candidate or make human-to-human connections.
And as long as AI can’t do that, well, talent still plays a key role.
Conclusion: What will AI-powered hiring look like in 2026?
AI advances are inevitable. But the role of recruiters and talent teams will be to transform the function from just filling jobs to strategic talent partnership that brings real benefit to the business.
It can be easy to feel either defeatist or all-in on AI adoption across HR, talent, and people functions. But we still have incredible chances to decide what good looks like, how we’re going to connect with candidates, and what a better, fairer, more human version of recruitment looks like in 2026.
At the end of 2026, the tools we’re talking about might have changed, but the question still won’t be about technology.
It’ll be about who took back the time AI created, and what they chose to do with it.
Still have questions about AI, recruitment, and all things talent? Join The Herd, Zinc’s community of forward-thinking HR and talent professionals, for a monthly dose of news and invites to exclusive in-person events.
