Everything now hinges on how well people and AI work together

AI use at work has skyrocketed in the last two years. Nearly nine in ten organisations now use AI regularly in at least one part of the business. But only 6% are getting significant value from it, and just 37% see any impact on the bottom line¹. So in most teams, the tools are there and people are using them, but few leaders can say whether the work is actually getting better.

Two years ago the question was what AI could do – how good is it really. Now, as the models have improved, the question becomes how we can use AI to the best of its ability and people and AI work side by side. Leaders need to be asking how the human+AI partnership is working and benefitting the organisation, or whether it’s just a claim.

Neither extreme should win

A business that hands the work over to the tools loses out on hard-won human expertise and intuition. Unless a company was founded to be run by AI with one or two human operators, it’s extremely difficult to pivot to a 100% AI operation. The challenge lies in working out what work to hand over to AI, and what work a human expert does.

In a survey of 1,150 US desk workers, 40% said they had received AI-made work in the past month from a colleague that looked fine at first, but had no real substance in it. On average, we lose two hours figuring that out and correcting the work². This is because AI tools are fast and confident whether or not they’re right, and a person who trusts it inherits that confidence.

AI tools are fast and confident whether or not they’re right, and a person who trusts it inherits that confidence.

But the opposite is also true. A business that keeps the tools at arm’s length also loses. Not thinking about how AI can help means missing out on growth, innovation, efficiency, or all three. And clients now assume AI is being used, so they expect to pay less for work they think a machine did. A business that cannot show what its people add has no answer to that.

That leaves one winning position: human and tool together, with the human responsible for the outcome.

Case study

Recruitment

The recruitment sector is a good example at the moment because the job depends on human judgement, and because the pressure has come from many angles all at once. They were quick to adopt automation for screening job applicants against the job criteria, but the EU AI Act now means the sector has to be extremely careful how they use AI when it comes to making decisions that could affect people’s lives and livelihoods.

The sector is experiencing other pressures like:

• Fees

  Clients are using AI as the reason to push fees from 15% towards 10%³.

• Clients

  Some hiring managers are asking recruitment agencies for interview notes and a human’s assessment instead of a shortlist.

• Candidates

  Three in ten UK job seekers have walked away from a hiring process because it used an AI interview⁴.

• The law

  Since 5th February 2026, UK law says a candidate must be told when a machine alone makes a big decision about them, such as rejecting them. They must be able to challenge it, and to get a person to look at it⁵.

In response to these pressures, five UK recruitment agencies (Prospectus, Carrington West, Aligra, Staff One and Harvey John) have published policies on how they use AI^{6, 7, 8}. The policies generally say: AI helps, but the human consultant decides. This is an interesting signal because it’s highly likely that as regulation becomes more widespread and more enforced, every sector will have to be transparent and say where the human sits inside their processes.

The law is asking every sector the same question

The EU AI Act affects any business operating in Europe, not just recruiters. Since February 2025, any organisation using AI in the EU has had to take steps to build AI literacy among its staff, in proportion to each person’s role. That includes UK firms whose AI work reaches EU clients. Since 2 August 2026, national regulators have been able to enforce it, and each country sets its own fines⁹. The stricter rules for high-risk uses, such as hiring, credit and education, have been pushed back to December 2027¹⁰. So for now the weight of the law falls on people, not systems. But the question is coming: can you show that your staff know how to work with AI?

Two things matter here. Literacy is knowing when to trust the tool and when to check it. Capability is doing better work with it. Neither is learned in a training room. Recruitment as a sector is not special here, but as an example, recruitment consultants learn on the job, with real candidates and real consequences, like most people in most jobs. AI literacy is learned the same way – working with it on live work to see not only what it can do, but how it works in a live environment alongside teams. After all, uses for AI are discovered, not invented.

Asking better questions about AI at work

A successful human+AI workflow means someone can exercise discernment and look at any piece of work and say what the tool did, what they checked, what they changed and why. Put simply, it’s what it looks like when someone uses AI well.

We’ve been using examples from the recruitment industry to think through this perspective, but swap the recruitment consultant for a lawyer, a marketer, an analyst or a teacher. The question is the same. It’s much more important for a leadership team to know how many people can answer that question about their own work than how many are using the tools.

How to know where you stand

Public promises and policies – and the PR that goes with them – don’t prove anything. Everyone has only just started to work out what their work life looks like with AI. Uses for AI are still being discovered, so the best thing to do if you lead a team is to find out where everyone is first, then make an action plan. Because AI moves so fast, it’s best to start somewhere and adjust as you go, rather than wait for a perfect decision.

In any organisation that has had the tools for a while, the human+AI partnership already exists in pockets: people who have worked out what the tool is good at and quietly built their work around it. The first job is to find them and know what they’re doing.