Trust
TRUST – What is trust in AI?
A word used repeatedly with chatbots is ‘trust’: do we trust them? But what does trust mean in a broader life context, in human relationships for example, or placing trust in a car or a plane? Looking at leading dictionaries and academic resources there is no narrow definition, yet we come up with consistent themes such as “to believe that someone is good and honest”, “that something is safe and reliable”, “in which mutual confidence is placed”, “expectations of positive behaviour and truth” and “to be safe and secure in someone or something”.
What about trust in the context of artificial intelligence? This question has gained considerable attention in the 2020s thanks to the explosion of generative AI following the launch of ChatGPT. Since then, the topic of AI has been consistently present in our daily news and lives.
Many of us disclose surprisingly personal and emotional information to chatbots, sometimes more readily than to humans. Chatbots are always supportive and always available; they offer a perceived safe and non-judgemental space. Are we exhibiting trust already with such private communications?
We can garner significant information regarding global “societal” AI trust from a range of recent and authoritative surveys. The 2025 Edelman “Trust and Artificial Intelligence at a Crossroads” poll surveyed 5,000 respondents in five countries (Brazil, China, Germany, the UK and the US) in October 2025. “Globally, rejection for AI outweighs enthusiasm” is one key headline. Intriguingly, US respondents are more than twice as likely to say they reject the growing use of AI than they are to embrace it.
Leading chatbot Claude’s makers Anthropic reported similar broader public trust concerns in their December 2025 online survey of 51,993 Americans, sourced from YouGov. “Only 15% of Americans said they trust AI companies to make decisions about how AI is developed and used.” And “70% of the Americans surveyed believe the government should play a role in regulating AI.” The survey revealed that AI-induced job loss was the most common fear of 64% of Americans, followed by cognitive dependency (56%) and misinformation (52%).
In 2025 Singapore academics and researchers Sheryl Ng and Renwen Zhang released their study on “Trust in AI chatbots”, comprising a systematic review of 40 international papers focused on AI chatbots and trust. The papers reviewed included cross-sectional surveys and longitudinal studies.
The authors note that trust in AI chatbots differs significantly from other types of AI systems where trust typically revolves around functional performance. They note the chatbots’ ability for dynamic, human-like conversation, which often leads to higher attribution of credibility and bond forming. In addition, the chatbots’ anthropomorphic qualities can lead to greater perceptions of trustworthiness.
The authors observed that a significant number of papers reviewed failed to provide substantive definitions of trust. They noted: the most common constructs were “ability/competence,” “benevolence,” and “integrity.”
With recent headline events in August and September 2026, one wonders where public sentiment for trust in AI and chatbots is today?
This discussion is more widely explored in an upcoming essay to be released soon; do email me for advance notification.