I recently sat in on an excellent session run by Teamshaper associate Tom Bryant, in front of a room full of policy professionals from across the civil service. At the very start, Tom asked us to choose which corner of the room to go to, where he had placed signs saying ‘excited’, ‘suspicious’, ‘fearful’ and ‘curious’. As I rushed to the ‘excited’ corner, I wasn’t surprised to see a fairly even split between the four.
This was a highly savvy audience at the top of their profession, people in charge of developing policy for our country in areas from housing to the environment, education to security. Within that group, a sizeable number were sprinting ahead on AI adoption and just as many were somewhat left behind, while others were using AI but uneasy about it, and some weren’t yet on the bandwagon but seemingly wanted to be.
The room was a microcosm of many workplaces across the public sector at the moment. AI has already changed many aspects of the work, and many departments now have AI adoption teams and projects running to remap processes and improve efficiency, at a pace of change that is dizzying.
You may say we’ve been through this before, with the service redesign rush of digitisation, and before that when the internet came along and changed public services and the way the public sector works. Those changes were led centrally, by digital transformation teams and IT departments, because they held the expertise and spoke the language of the technology. The main difference this time is that a central team is no longer the only route to change, because AI removes the need to be proficient in the language. You no longer need to know code or use any particular piece of software to create most kinds of apps, so we can all cut out the middlemen, from the transformation professionals and IT consultants to the graphic designer or the UX expert, as AI can do all those things for us.
This means the people on the ground, who know the work best, can create the applications themselves. Policy people can create an agent that replies to parliamentary questions without needing a tech person to make it happen, and who better to create that agent than the people who have always answered PQs? Customer experience professionals can build digital services, because who better to design them than the people who engage with service users day in day out, without a middleman adding friction, complexity and cost?
This is all wonderfully liberating, but the middlemen brought more than technical skill. They brought checks and balances, and they brought prior knowledge in areas such as process design, user interface, and maintenance and update logs, all of which help produce a better solution, whatever the underlying issue the application is there to address or support. They also slowed things down in a positive way, so that problems and red flags became visible earlier and could be addressed systematically. With that layer gone, each person now meets AI on their own terms, and from an organisational development perspective this raises a red flag in each corner of Tom’s room, which as a training provider we’re increasingly focused on.
1. The excited: some people get carried away
With nobody standing between the idea and the finished product, those in the ‘excited’ corner, like me, can become so reliant on AI that they open the organisation up to risk, create errors and lose trust through work that is under par. I heard of a leader who built an agent to deal with his inbox entirely independently of him, to the point where AI was replying to his emails. Needless to say, his team was not impressed, a lot of nonsense came out of it and there were a couple of seriously embarrassing incidents. Some public sector organisations have used AI to create content that is terrible and looks like AI, which causes people who notice it to lose trust and respect. A friend of mine asked his AI to look for options for a branded product for his business, gave it free rein, and one day received three samples from China in the post that the AI had ordered entirely on its own.
I’m sure you’ve come across examples of your own. This is a very real problem, particularly in a heavily regulated environment such as the public sector, where the main currency is public trust, which is so easily ruined and so hard to rebuild.
2. The fearful: some people are full of dread
When a change was delivered by a project team, everyone moved with it, but when each person builds for themselves, those who don’t build fall further behind. These are the people who are completely off the bandwagon, or at best have occasionally used the free version of ChatGPT to write birthday cards. They may always have been late adopters, or there may be other reasons, but the fact is they aren’t taking part, and this creates serious organisational issues. When some staff are developing their own agents and others aren’t even using basic functionality, you get a two-tier workforce that differs in the way it works and, over time, in pace of work, level of error, compliance and tone. Take proofreading, minute-taking or checking against compliance regulations, tasks that take an AI user minutes and a non-user hours. This is very unhealthy for any workforce, and the gaps will only grow over time if not addressed, until they’re seen as performance issues.
3. The curious: willing, but blocked
The curious want to engage with the main event, but with no expert leading them in, they’re facing blockers of some kind. They can be brought on board more easily than the fearful, but it takes understanding what the blockers are and addressing them, both systematically and individually. These are people who may need more than the tools and the confidence: they may need more handholding than others, they may learn differently or they may face other blockers altogether. Without that help close to their work, they are failed by the L&D team that sends them to a generic 90-minute online course, which expects them to do much of the work of translating what they learn into their daily work realities by themselves.
4. The suspicious: using AI, but uneasy
With no technical gatekeeper to vouch for what is safe, the suspicious are left to judge the risk for themselves. They’re most likely using AI, perhaps even extensively, but they’re very uneasy about it. They’ve read the articles about the end of the world, the loss of judgement and the security risks, and they actively try to limit their use, suspecting that if they share something wrongly it may end up in the public domain. They aren’t late adopters at all, they’re fearful of this particular technology, and by all accounts for pretty good reasons, given what we hear from the AI companies themselves. As a result, they underuse AI or share too little with it, or when they delegate to it, they check it so meticulously that the time saving evaporates, or worse.
Some of the fear is misguided, and some is justified and should occupy all of us, but the truth is in the middle and this group has gone too far, in the same way some of the excited have. Factual detail about what the risk is will help them, and at the same time, strong guardrails that mitigate the risk in a visible and above-board way are in everyone’s interest. Working at team or organisational level also helps, because they feel the safety in numbers, and the organisational knowledge and backing to manage the risk in a way that is clear to everyone. These are often people simply looking for guidelines and boundaries, who feel the current environment favours the mavericks, and that makes them uneasy.
Where our approach comes in
If the middlemen have gone, the checks, support and structure they brought still need a home. Central AI adoption and digital transformation teams will continue to do their work, and we as trainers focus on the teams on the ground, which is what gets the central team the engagement that will ultimately determine its success. We go beyond the GCSE prompting framework (goal, context, source, expectations) to building projects, agents and advanced process redesign, while keeping a firm eye on the human side of adoption, which is different from and more profound than any previous technological change in the workplace. We support individuals to adapt, overcome fear, feel in control and keep up, and at team and organisational level we create the structures that make each team or organisation as strong as its fastest adopter, with no one left behind.
We build processes with your team so that every team member benefits from the ingenuity of those who run ahead, and we make sure the right safeguards are factored into the process, workflow or service design, so individuals can trust what they’ve built and those with fears can rationalise them and understand clearly what risk they face and how to mitigate it.
Have a look at what we do in this area. It can make a major difference to your AI adoption journey, both in how a diverse workforce goes through it together and in designing the new processes and workflows the right way, joined up and safe.
