Trust in AI

Trust & AI

Imagine intelligence is woven into nearly every surface around us. When you look at these surfaces, they look back, detecting and reacting to your presence. Autonomous robots have taken over routine chores like cleaning and cooking, boosting productivity and efficiency. Self-driving cars dominate the roads, easing congestion and reducing road accidents. Virtual reality and augmented reality have revolutionized education, entertainment, and communication. Personal digital assistants help plan daily activities, monitor health, and notify users about diet, stress and health issues. This is the promise of AI in 2030.

Despite the advancements in AI, the widespread adoption of these technologies’ hinges on one critical factor: trust.

Why is trust essential

Trust provides assurance in the consistency, accuracy and integrity of technology. ?Trust is needed for mass roll out and consumption of technology. The process is no different than introduction of new drugs, vaccines or even electronic equipment example TV, mobile phone. In every case, the product undergoes a lot of testing to evidence that they are safe for public use.

As a species our success is primarily attributed to its intelligence and cooperation generated from trust in large numbers - in nation states million and billion people strong and internationally among nation states for common good. Trust enables people to group beyond family lines into villages, cities, nation states. Making governments and following laws - even with complete strangers, trusting they would do the same. Trust also enables division of labour, specialising in skills and trade - the basis of modern economics.

For AI & Robots, its physical manifestation, to actively engage and work with people - it will have to earn this trust. Example people trust each other to follow traffic rules and will not automatically trust a self-driving vehicle to consistently perform competently as their human counterparts. Hence, trust will be a prerequisite for technology to be introduced enmass.

Current state of trust in Technology

In the digital era, different industries adopted to different levels of trust, determining the levels based on customers attitude and regulatory requirements. Example most users are aware and accept social media (example Facebook, Instagram), Search tools (example Google), and smart home devices (example Google home and Alexa) monetise their information to provide the service. But the same users do not accept unauthorised financial transactions or an unplanned system outage in Banks or Financial services, like Barclays Payday system outage on 31st January 2025.

The regulatory framework evolved around the customer expectations and classified financial services under heavy regulations, in comparison to other industries. This led to Banking and financial industry adapt a risk averse approach, providing a higher level of trust in its system. And some organisation like Visa Europe imbibed their ambition around trust - "to be the most trusted currency", creating a bottom-up push.

In comparison, other industries, again reflecting on customers attitude and the corresponding regulatory obligations had a light touch approach to trust in IT systems. This led to the introduction of new regulations including data protection & Network and Information Systems Regulations 2018 by the Information commissioner’s office (ICO), with varying levels of success.

The different approaches to trust were reflected in the 2018 TSB bank and British Airways IT incidents, with very different outcomes.

·?????? In the TSB bank incident, many customers were locked out of their accounts, some customers received access to the confidential records of others whereas some experienced unauthorised transactions from their bank account. The problems continued for many weeks and TSB came under fierce criticism for the IT failings. TSB was penalised a total of ￡81.35M:

o?? The Financial Conduct Authority (FCA) - ￡29.75M,

o?? Bank of England Prudential Regulation Authority (PRA) - ￡18.9 ??

o?? and financial redressal to 5.2M customers - ￡32.7M .

o?? In addition, TSB CEO Paul Pester had to resign.

·?????? In comparison, the British airways incident had led to loss of customer including log in, payment card and travel booking details as well name and address information for 400K customers. BA was penalised ￡20M by ICO for the incident.

Future of AI Trust

Unlike Digital trust, AI trust will be industry agnostic and will require similar level of trusts from both Financial and other industries. This is because AI will be as ubiquitous as electricity and like electricity, which is used for myriad activities from lighting, cooking, cleaning, entertainment and even running automobiles - AI will be used for all those activities by adding a layer of intelligence to them, making them more efficient and valuable. As a result, like Electricity, AI will be covered by regulations agonistic to the industry consuming it - Financial or non-financial.

The implications of this transition will be immense and industries which have bottom-up trust i.e. already have systems and processes that generate and monitor trust in systems will be able to adapt AI seamlessly whereas other industries, which had taken a minimalistic approach to comply with regulations, will have a steep maturity curve to climb changing organisation cultures and introducing systems and processes.

Security & Trust

Security controls generate trust by evidencing system performance and providing assurance and oversight of AI systems, bringing the risk down to tolerable levels. And because AI systems are fuelled by the data, Security controls monitor the usage and access to the data – alerting any misuse.

Security systems and processes will play a key role in generating trust and achieving the benefits from AI / Robotics while keeping the risk to a manageable level. The challenge however will remain to achieve the target state efficiently and without time and cost overruns.

Summary

There is an element of inevitability in AI extending the benefits introduced by digitalisation, but for mass consumption – it will have to generate trust among its users. I agree with Bruce Schneier summary on the importance of Trust “it doesn’t matter how big your neocortex is or how abstractly you can reason: unless you can trust others, your species will forever remain stuck in the Stone age”.

What are your thoughts on Trust in AI? I would love to hear your ideas.

Thanks for reading