Simon Glairy is a recognized expert in the fields of insurance and Insurtech, with a specialized focus on risk management and AI-driven risk assessment. As algorithms increasingly take the wheel in underwriting and claims processing, Simon provides a vital voice on the “human element” that must remain at the core of the industry. His work often bridges the gap between high-level technological innovation and the practical, ethical governance required to maintain public trust. In this discussion, we explore the findings of a critical new report on artificial intelligence and why technical access does not always equate to professional competence.
The following conversation examines the growing disconnect between the rapid adoption of AI tools and the actual “fluency” of the professionals tasked with managing them. We delve into the risks of a “wait-and-see” regulatory environment and the specific challenges faced by boards and frontline staff who may feel pressured to prioritize efficiency over sound judgment. Furthermore, the discussion highlights the specific data points regarding project failures and the practical steps brokers should take to ensure their carrier partners are offering more than just a nominal “human in the loop.”
While many insurance professionals are learning to operate AI tools, there is a significant difference between basic usage and what the CII calls “AI fluency.” How would you describe the danger of deploying these technologies before a team truly understands how to challenge them?
The danger lies in a deceptive sense of security where firms believe they are innovating while they are actually just automating risks they don’t fully comprehend. When we talk about the “fluency gap,” we are looking at a scenario where 75% of UK financial services firms are already using AI, yet many of the people at the controls are merely clicking through interfaces out of obligation rather than insight. If a staff member lacks the professional skepticism to question an automated output, the AI essentially becomes the final authority, diluting the ethical standards that underpin our entire profession. We saw this concern echoed during the CII’s roundtable in June, where experts noted that without deep training, firms might be driven by a frantic fear of falling behind competitors rather than a clear sense of purpose. This results in a “compliance-only” culture where the tool is used, but the human judgment required to spot a wrong or unsuitable decision for a client is completely absent.
The report suggests that “human in the loop” oversight is often treated as a formality rather than a genuine safeguard. In your view, what does active and informed accountability look like in a real-world insurance setting?
True accountability is a much more demanding standard than simply having a person give a nominal “thumbs up” to an AI’s recommendation before it reaches a customer. It requires that the individual in the loop is active, informed, and has the genuine authority to overrule the machine without fear of repercussions regarding “efficiency metrics.” For a person to be a real safeguard, they need visibility into how the model reached its conclusion and the specific training to recognize when a decision is fundamentally wrong for a client’s unique circumstances. We have to move away from the idea that a human presence is an automatic safety net; if that person is just a rubber stamp, they aren’t managing risk, they are just participating in its execution. Active oversight means the human is the navigator, not just a passenger watching the autopilot take the ship toward the rocks.
With the House of Commons Treasury Committee criticizing regulators for a “wait-and-see” approach, how are firms currently navigating the gap between rapid deployment and delayed official guidance?
Firms are currently operating in a high-stakes vacuum, where the professional judgment of their staff has become the primary safeguard in the absence of detailed rules from the FCA or the Bank of England. The Treasury Committee’s conclusion in January 2026 was quite sobering, noting that the “wait-and-see” stance could expose consumers and the entire financial system to serious harm. While the FCA is expected to publish practical guidance on consumer protection and senior manager accountability by the end of 2026, firms cannot afford to sit on their hands until then. This regulatory gap is precisely why internal governance and staff fluency are so vital right now; the industry’s own ethical framework must act as the bridge. If a company moves ahead of the regulations without building internal capability, they are essentially betting their reputation on a technological “black box” that no one in the building knows how to open.
The Grant Thornton survey of 100 insurance executives found that only 24% are very confident they could pass an independent AI governance review within 90 days. Why is there such a massive disconnect between having a framework on paper and being ready for a real audit?
That 24% figure is a startling wake-up call because it reveals that while many firms have the “paperwork” of governance, they lack the operational reality of it. The same survey pointed out that 44% of AI projects have already failed or underperformed due to governance and compliance challenges, which suggests that these frameworks aren’t actually guiding the technology effectively. It is one thing to have a policy document sitting in a digital folder, but it is quite another to have a frontline where 39% of employees are identified as needing significantly more support to work with AI. When only a quarter of executives feel they can stand up to scrutiny, it tells us that the “fluency gap” has reached the highest levels of leadership. The governance is outpacing the people, creating a fragile structure that looks stable from the outside but lacks the internal expertise to withstand a rigorous investigation.
The Lloyd’s market shows a high percentage of firms with AI frameworks in place, yet deployment is mostly seen in operational efficiency. What does this tell us about the industry’s current comfort level with using AI for underwriting or claims?
The data from the Lloyd’s market—where 72% of firms have frameworks and another 21% are developing them—shows a clear hesitation to let AI handle the “heavy lifting” of the insurance craft. By concentrating AI in operational efficiency rather than the nuanced worlds of underwriting or claims decisions, the industry is signaling that it doesn’t yet trust the technology, or its people, to manage complex risks. This concentration suggests that governance structures are often a protective shell designed to satisfy stakeholders, while the actual capability to use AI for high-stakes decisions remains underdeveloped. There is a palpable tension here: firms want the cost-saving benefits of automation, but they are terrified of the liability that comes with an automated underwriting error. Until we bridge the talent and upskilling gap, which 29% of executives cite as a top barrier, AI will remain a tool for moving files faster rather than making better decisions.
For brokers who sit between the client and the carrier, what specific questions should they be asking their partners to ensure that AI-driven decisions are being handled responsibly?
Brokers have a direct stake in this because they are the ones who must stand behind a placement or a claims outcome when a client comes knocking. The most important question a broker can ask a carrier or MGA is not “Do you have a human in the loop?” but rather “Is the person reviewing this output equipped and empowered to overrule it?” You want to know if the reviewer has actual visibility into the model’s logic and if the carrier can demonstrate a track record of humans correcting automated errors. If a carrier cannot answer these questions with confidence, the broker should be very concerned that the “human in the loop” is just a formality. A broker’s role is to protect the client, and that requires ensuring that the carrier’s AI isn’t just a machine producing efficient but ultimately flawed or unfair results.
Given the current trajectory of technology and the slow arrival of regulatory guidance, what is your forecast for the insurance industry over the next twelve months?
I believe the next twelve months will be a period of “forced maturity” where the gap between having AI governance on paper and having genuinely capable staff will become the ultimate test for the industry. We are going to see a sharp divide: some firms will double down on continuous professional development and maturity models, while others will continue to “click through” until a high-profile failure or a failed audit forces their hand. With the FCA’s guidance not arriving until the end of 2026, the firms that thrive will be those that treat AI fluency as a core professional skill rather than a technical add-on. We will likely see more AI projects stall or be pulled back as the reality of the 44% failure rate sinks in, leading to a much-needed shift from “AI at all costs” to “AI with purpose.” Ultimately, the industry will realize that no amount of code can replace the human judgment and ethical standards that are the foundation of insurance trust.
