“Is it cheating?”
That’s just one of the intriguing cultural issues sitting behind the UK’s skills challenge on AI, as we heard in our latest BusinessLDN and Deloitte AI SteerCo, which met again this week. Our task this time out? To look at what AI means for future skill requirements and what changes are required to train tomorrow’s workforce.
We started by checking in on what’s stayed the same and what’s changed since we last met in June.
The labour market remains complex: simultaneously heating up and cooling down. Heating up as the scramble for AI talent continues, fuelled by the major AI players piling into the London market. And yet cooling down, as entry-level roles drop off in some cases for a range of reasons – including but not limited to AI – although it should be noted that many firms are maintaining their intake.
Firms are continuing to find their groove on training, with a continued emphasis on peer-led learning and team-based training.
And the emphasis on finding the right risk-reward balance continues, with firms balancing efficiency gains with security risks and return on investment questions.
So, what’s new?
Training continues to evolve. The focus on in-person, practical training is further dialled up, with greater attention paid to how to embed and use the tools, and to encouraging people to think about how AI can help the flow of their work. There’s also more emphasis on critical-thinking training, including how to interpret data.
AI capability is showing up more in interviews, with many candidates now talking intuitively about how they would apply a range of AI tools to problem-solve different scenarios.
As the roll-out of AI becomes more mature, many firms are now allocating costs to business units rather than being met from a centralised IT budget. This is putting more scrutiny on use cases along with the costs of licenses and tokens as managers take a closer look at returns on investment.
Linked to this, firms are continuing to grapple with how to bank some of the efficiency savings. With time savings scattered around individual tasks, there’s a danger the efficiencies “get procrastinated away”. The solution is to frame it around time saved on a process, not time saved on a person.
More thought is going into getting the blend of training right. With only so much “training bandwidth”, if you put all the emphasis on ethics and compliance, you miss the crucial bit of helping people learn how to get the best out of the tools. There’s also much more thought going into targeting who needs what training. Organisations are realising they don’t need to get everyone to the top of the curve. It can be a wasted effort to get everyone capable of building agents – far more important is that agents are well deployed and widely used.
Managers and HR are grappling with a 21st-century version of “keeping up with the Joneses” as colleagues are questioning why some colleagues have access to Claude or more premium versions of AI tools than they do, rather than focusing on what they need to do their job well.
And there is a growing awareness and focus on the use of “shadow AI”, where people are going off the company books to use AI on their personal devices or with personal logins, giving governance, IT and chief risk officers the heebie-jeebies.
And the answer to the cheating question? It’s something that is running from how young people in school are being taught through to senior managers in business cautious about what their teams will think. The answer? Think about it as “How do I use AI to support me in my work, not do my work for me?”
P.S. Look out for Deloitte’s Gen AI workforce survey released next week. We were treated to a sneak preview. The survey’s breadth and depth offer fascinating insights into how workers are using AI. Look out for some interesting contrasts and similarities between London and the UK. And how, when it comes down to it, we’re a nation of dabblers…