The AI Muscle: Building Resilience in an Uncertain Landscape

I recently had the pleasure of attending a roundtable discussion hosted by Neueda, joined by a group of HR, Learning and Development, and tech leaders from the UK’s banking, insurance, healthcare, and financial services sectors. The goal was to explore the people-centred challenges of a fast-evolving AI landscape. 

The atmosphere was one of palpable urgency, which reminded me to an extent of the early days of the Internet – although this time the stakes are significantly higher. This group were collectively further ahead in AI adoption compared to other sectors I know – where the conversation can still be driven by an element of FOMO (fear of missing out) – and it was clear all were tackling the fundamental question: how do we prepare our people for an AI future? 

Building the “AI muscle”

One of the most significant challenges discussed was the breakdown of traditional strategic planning. In an environment moving at this speed, planning even six to nine months ahead feels like a formidable task. As one attendee put it: stop trying to plan for the long horizon and focus on building the ‘AI muscle’. 

This means developing an organisational capability to adapt, regardless of how the specific tools change. Organisations are already moving beyond simple chatbots and are trialling agentic systems. Interestingly, the speed of MVP (Minimum Viable Product) development has changed the financial calculus; leaders can now justify starting speculative projects from scratch because the cost vs. risk of failure calculation has become acceptable. 

The Broken Ladder and Cognitive Offloading 

During the session, I alluded to my own ‘Broken Ladder’ research in the professional services sector. Whist the roundtable attendees came from corporate structures rather than partnership-based, billable-hour firms, the parallels were significant. 

Across the board, organisations and individuals are coming to terms with this new, AI-driven era and the path ahead may not be a smooth one. The consensus I’ve found is that the technology is beyond doubt here to stay and will only become more prevalent, but what exercises all minds is keeping up with the sheer pace of change.   

Amongst others, a serious concern is the risk of cognitive offloading. As AI handles more entry-level tasks, we risk removing the rungs of the ladder that allow junior staff to build foundational expertise. If the grunt work has gone, how do we ensure the next generation develops the deep intuition required for senior leadership? 

In this new world, specific AI competency is becoming table stakes – the bare minimum for entry. The true differentiators for the workforce are critical thinking, curiosity and adaptability.  

The attrition shift

Contrary to the common fear of mass layoffs, the organisations and institutions represented were not actively reducing headcount. However, they aren’t aggressively recruiting either. Instead, we are entering a phase of natural attrition. As the workplace evolves, those resistant to change or unwilling to adapt are moving on, allowing for a gradual, organic shift in the workforce’s collective skill set. 

To support those who stay, organisations need AI champions – internal advocates who can bridge the gap between technical potential and daily practice. But these champions aren’t enough on their own. Leadership must provide the time and space for people to learn properly, rather than simply layering new tools on top of an already full workload. 

ROI, Trust and Integrity 

As all sectors move beyond pilot schemes there is inevitable pressure to demonstrate measurable ROI. Enterprise AI tools aren’t cheap and need to pay their way, yet identifying and agreeing on appropriate metrics is not straightforward. Hard objective data (for example. speed of turnaround) may be one thing, but there will be softer, subjective considerations too. Staff satisfaction could be one. Are the tools making their lives easier, or actually adding to the load? 

In sectors like healthcare, collecting such data is further complicated by the need for absolute integrity in training data and watertight audit trails. 

There was a candid discussion about the need for a ‘human-in-the-loop’. While often cited as a safety feature, there were concerns about human fallibility. Humans get tired, they get distracted, and they are prone to bias. The group suggested that the human-in-the-loop is being maintained primarily for liability; to ensure there is a person to blame when things go wrong (I’ve heard this from lawyers too). This raises a critical question for the near future: how well are regulatory bodies and insurers keeping up with the reality of professional liability in an AI-driven age? 

Help or Hindrance?

A point of discussion I often raise is focused on the changing relationship between experts and their clients or colleagues. We are increasingly seeing situations where a client uses AI to research a matter before even approaching the human expert, often to reduce the cost of advice, however measured. 

Does this help by providing a more informed starting point, or does it hinder the process by forcing the expert to ‘un-teach’ AI-generated hallucinations or oversimplifications? In highly regulated industries, this ‘un-teaching’ could well carry much greater weight. When a client or colleague presents an AI-generated briefing note that inadvertently misinterprets say FCA guidance or a complex risk framework, the expert’s role shifts from adviser to critical safeguard. 

In these cases, correcting a hallucination isn’t just about accuracy – it’s about maintaining the integrity of audit trails and meeting the rigorous standards expected by shareholders and regulatory bodies alike. Ironically it could add to time spent, too.  

Whether we are answerable to external clients or internal stakeholders and shareholders, the pressure to prove value remains intense. 

Final Reflections

The Neueda roundtable made it clear that while the technology is in every sense fast, the human element is where the most complex work remains. We are all feeling our way through this transition. The winners won’t necessarily be the ones with the best software, but the ones who have built the strongest AI muscle and given their people the space to grow alongside the machines. 


About the author

James Tuke is the founder of multiple tech-related ventures, such as Treat Digital, and is now also the CEO of the AI Futures Forum, a site dedicated to in-depth discussion about AI, and writer of “A Short Walk in AI”. He also advises professional services firms and senior management about their AI strategy, with a particular focus on people-centred planning.