Are humans ready to engage with bots?

As expectations on digital user-experience rise, and pressure from non-traditional players grows, insurers are turning towards AI-enabled technology for customer interaction. While this promises personalised, fast and intuitive end-to-end insurance experience, the industry needs to beware of regulations relating to the use of AI, the limitations of the technologies, and acknowledge that many clients still want human interaction.

Customer expectations on digital user experience are rising. To meet expectations of high-value and frequent-touch interactions, more personalized services and 24/7 support, the insurance industry is turning to digital intermediaries and AI-enabled technologies for customer-facing activities.1 This is creating a new dimension in client-machine interactions.

Developments in natural language processing, computer vision including image recognition and other technologies bring many opportunities. Virtual agents like chatbots allow clients to address their issues at any time. Robots can also send reminders to clients to finish processes if they were interrupted, request missing information and analyze images of incidents. AI-enabled assistants can sift through vast amounts of client data and formulate more personalized service much faster than a human could. AI can also help accelerate claims processing and underwriting services. Algorithms integrated with internal policy data can enable insurers to fully automate and execute claims payout in real time.2

AI has limits though, not least the lack of certain cognitive skills.3 This is one reason why humans are still needed, to take care of cases that require special handling. Machine learning technologies can spot unconventional cases, which should be then be referred to humans. Integrating all automated systems, and also making them work with humans to offer a reliable and efficient end-to-end customer experience can be challenging, and costly if not executed carefully.

There are other challenges also. For instance, virtual assistants can carry out a variety of tasks, but these may be regulated differently across various jurisdictions. In a recent case in the UK, an insurer was fined for over-reliance on voice-analytics software, which led to some claims being unfairly declined or not adequately investigated.4 Large-scale serial claims could also arise should chat-bot software be incorrectly programmed. And should a claim be incorrectly accepted, this could create a precedent for similar claims. With respect to compliance or operations, the insurance industry’s use of virtual advice could also be exposed to risk arising from tighter regulatory restrictions.

An underlying consideration is that only 20% of consumers say they are comfortable using chatbots for their financial dealings.5 Tech-savvy generations are more accepting but overall, many are not comfortable with the idea of taking advice from a robot. While expectations on digital user-experience are rising, customer reluctance as well as risks related to regulatory restrictions and technological errors need to be balanced when deciding where and to what extent to go artificial and automatic.


1. The Age of With – Accelerating the impact of augmented intelligence in insurance, Deloitte, 2020, age-of-with-insurance.html
2. Ibid.
3. “An understanding of AI’s limitations is starting to sink in”, The Economist, 13 June 2020, understanding-of-ais-limitations-is-starting-to-sink-in
4. J. P. Raman and R. Lam, Artificial intelligence applications in financial systems, Oliver Wyman, 2019, artificial-intelligence-applications-in-financial-services.html
5. Acceptance of artificial intelligence chatbots by customer worldwide, as of 2017, by service, Statista, customer-chatbot-acceptance-by-industry/


Related SONAR content

SONAR 2021: New emerging risk insights

While the COVID-19 crisis dominates the risk landscape, other emerging risks and trends arise from developments in human-machine interaction, connected infrastructures, and ethics and sustainability.