With an estimated 11 million people in the UK now using generative AI at work — often informally, ‘side of desk’ — employers face a growing challenge: how do they harness AI’s transformative productivity potential without compromising trust, safety or wellbeing?
And, given the risks of everything from “hallucinating robots” to rising employee fear and data leaks, are employers doing enough?
We spoke to Deloitte’s UK Trustworthy AI Lead Lewis Keating about the real risks, the overlooked human factors, and why writing a policy isn’t enough.
Your job title is ‘UK Trustworthy AI lead’. What does that mean?
As Deloitte focuses on technology and business transformation, specifically AI, I advise organisations on how to use AI in a way which is safe, secure and ethical. And, increasingly, in compliance with the range of evolving regulations.
Trust is fundamental.
CIPD Chief Peter Cheese has spoken to us before about his concerns that employers aren’t doing enough to plan for AI. Do you have any research, or a gut feel, for whether his concerns are warranted?
It is now estimated that 11 million people in the UK are using Gen AI for work, and my experience tells me that many organisations don’t yet have approved tools for their people to use. I share the concerns about employees using it informally ‘side of desk’.
Why are you concerned about employees using it informally ‘side of desk’?
Well, for one, AI ‘hallucinates’ which is the term for when the AI platform makes things up.
For instance, when lawyers have used it, it’s responded with completely made-up case law examples.
Because AI is trained on huge amounts of data, it’s very good at sounding confident and giving plausible sounding responses. However, they’re not always accurate due to the nature of how AI generates its content.
Another concern for employers is data privacy, and how employees might use confidential company data with these Gen AI tools.
Do you think these glitches will ever get ironed out to the extent that we don’t need humans guiding AI at all?
History shows that every major technological shift has ultimately created more opportunities than it displaced.
AI won’t just change roles, it will essentially change how work gets done and create the opportunities to change the role that humans play.
This efficiency is further amplified by AI agents or Agentic AI.
What is agentic AI?
In simple terms, it’s AI with ‘agency’; it can make decisions on its own and take action, removing the need for a human to be in the loop.
A relatable example is asking AI to sort a user’s travel. It could book the hotel and train, without needing to consult a human. But the removal of a human comes with an increased risk, and organisations need to be thinking about how they manage this risk.
What are some of the biggest risks you’re seeing in relation to AI?
Employees perception of AI in the workplace is often: ‘I’m scared of this’ and ‘I don’t understand this’ or ‘is this going to take my job?’
Employers can underestimate the time and training needed to help humans adapt to using AI successfully and correctly. The opportunities are fantastic but changing human behaviour takes time and careful consideration.
What do you see as the main benefits of AI to the workplace?
From an economic perspective, this technology has the opportunity to arguably solve the UK’s productivity challenges.
Research shows between a 10% – 30% productivity increase if you train people on how to use AI properly.
Every employer has the opportunity to do what they do, better, and even reimagine what they do.
But the main barrier right now is trust. Our research shows that only 50% of people in the UK trust businesses and organisations to use AI tools responsibly.
When employees feel confident using this technology, and trust it, organisations will reap the benefits of AI.
What about risks associated specifically with Health and Wellbeing?
As with all technology available to a consumer, there’s risks around using AI tools. For example the free tools offering counselling services to talk to someone in a therapeutic way,may not be tailored correctly for that purpose, and users should be cautious.
What time scale do you think we are looking at to see this kind of transformative change where AI is embedded and we’re reaping these benefits?
Organisations are seeing benefits from AI now, and these benefits will continue to come as the technology evolves.
We spoke to Tolulope Oke, former Head of Inclusion & Diversity at Diageo, about AI and she’s concerned that there’s not enough talk from corporates on the effect on less developed countries and the environmental impact. What do you think?
A year ago, the environmental impact wasn’t a major part of the conversation about AI and its risks. Now it comes up in almost all of the work I do with organisations, and this is a positive change.
If a reader is reading this wondering where to start with its AI policy – what would you advise?
Firstly organisations should prioritise having an acceptable usage policy to give guidance on how, why and where to appropriately use AI in the workplace.
But it’s not enough to just do a policy – it needs to be accompanied by proper AI fluency training to really educate employees on what this technology is and how they can use it responsibly.
Secondly, think about how you are going to use AI in a trustworthy manner. For example, creating a steering committee with a wide range of expertise will ensure AI governance is done well. Think data, technology, privacy, security, HR, legal, compliance, business operations experts as it needs a broad spectrum of views coming together because this technology is complex and going to be used in a multitude of ways across all business functions.
What about gig workers – should they be included in an employer’s AI policy?
Yes – employers need to be giving clarity to all the people in their network. That might be slightly different for people who are freelancers, or working for you via a third party organisation, but they should absolutely be given clarity, both in policy training and contractual agreements.
How do employees keep themselves relevant in an AI world?
Embrace the technology, upskill themselves and learn about it so they know how to use it in their job.
I personally think there will always be the need for the personal human touch. Just look at what has happened with chess. There are AIs that are way better than humans at chess and, if you put two of these together, you will have a higher quality game of chess than ever.
However, no one watches it. The World Chess Championships is still between two humans. And, guess what? Thousands of people watch this. It’s real. It’s exciting. If we can’t interact as humans, where does that leave us?
You might also like:
