The Human in the Machine: A Plan for Trustworthy Artificial Intelligence — The University of British Columbia Okanagan (UBCO) researcher Madeleine Ransom says the path to trustworthy AI starts with realizing human expertise is still our most valuable resource

Date published: August 27, 2026 | Estimated read time: 5 min Social SCiences and Humanities Research Council

Legislators and stakeholders are working at full speed to develop policy and infrastructure to make our nation more “AI ready,” but a significant number of Canadians seems more interested in putting the brakes on AI.

A Leger survey from August 2025 found more than one-third of people surveyed predicted AI would be harmful to society. The rest were split (nearly evenly) between optimism and uncertainty. The ambivalence is pronounced.

“A lot of people in North America think AI is going to make things worse,” says Madeleine Ransom, Canada Research Chair in Artificial Intelligence, Well-Being and Ethics. “As a result, there’s been a lot of backlash and resistance.”

Since the public launch of generative AI platforms, beginning with ChatGPT in November 2022, people have raised concerns the world over. Not all reactions, though, have been as polarized as the ones in North America.         

“Polls show people tend to be more optimistic about AI in countries with stronger regulations and protections for people,” says Ransom, who is also an associate professor in UBCO’s Department of Economics, Philosophy and Political Science.

Those protections, which involve mitigating potential risks and harms such as job displacement, threats to privacy, and misinformation, are sometimes criticized on the grounds that oversight could curtail innovation. That may be true, but laissez-faire approaches have been shown to have limitations too. 

“Public trust diminishes when you have too little regulation,” says Ransom. “Most major tech corporations recognize this, which is why they’re actively calling for regulations and safety guidelines.”

In humanity we trust

Madeleine Ransom.

Photo: UBC Studios Okanagan

Public mistrust has inspired a “trustworthy AI” movement meant to address existing problems, such as “AI hallucinations” (literally making answers up and presenting them as fact), as well as other forms of misinformation. Another major concern involves biases that stem from AI having been trained with tainted data sets. These biases can lead to real-world discriminatory practices, such as the distinct pattern of gender bias found in financial institutions that use AI to evaluate credit applications. AI bias has been discovered in health care, justice and human resources, among other fields.

“Ideally, we want AI to catch and correct human biases rather than reinforce them,” says Ransom, noting that when users discover bias in AI systems they, quite understandably, begin to mistrust the technology altogether.

Ransom says she’s inspired by the work on trust being done by some of her graduate students, such as Bukunmi Okoduwa, a UBCO researcher originally from Nigeria who’s applying Ubuntu philosophy to AI, and bringing fresh perspectives to the field. That research all began with Ransom’s interest in something Okoduwa had written on Ubuntu and justice.

Ransom asked her if she thought there was a way to apply Ubuntu philosophy to trustworthy AI.

“At first, I thought, well, how is that going to be possible? But then I realized that AI often excludes minorities,” says Okoduwa. “Ubuntu theories of trust, though, involve the whole community and, if you bring the larger community into AI, that will reduce the structural biases and structural prejudices.”

Ransom says, in general, she thinks more holistic approaches to trustworthy AI, similar to the one Okoduwa is working on, can help us find a way to work towards a future that includes AI but, at the same time, protects privacy, mental health and employment for humanity.

“Building AI that is truly trustworthy means that we need to protect people from risks and harms,” says Ransom. “Ultimately, I think that building trustworthy AI is not just about designing a technology, it’s about designing a society.”

AI literacy for all

Before we can design a trustworthy society, though, Ransom says Canadians need to have a better grasp of AI basics, including ethical issues, detecting AI errors, and using these tools safely.

Madeleine Ransom in conversation with Bukunmi Okoduwa.

Photo: UBC Studios Okanagan

To that end, Ransom and her colleagues at UBCO’s Digital Transparency Research Excellence Cluster are working on building new AI literacy tools for Canadians, a project supported by a recent SSHRC Insight Grant. In addition, they’ve been doing many public-facing events, including an AI empowerment initiative run in partnership with the UBC Centre for Migration Studies that, last fall, saw Ransom and colleagues offer free AI literacy workshops for immigrants.

Ransom also works on outreach to produce public talks, panels, and townhall-style debates on AI ethics.

“One of the things that I argued recently at a public talk I gave in Vancouver is that, if we want a functional society worth living in, we can’t have that postwork future that some people think we’re headed towards,” she explains.

It’s not enough to just have some humans “in the loop,” she adds. In her research, she argues trustworthy AI is only attainable if we also have robust systems of checks and balances operated by communities of qualified experts.

Says Ransom, “Even if we assume people receive some sort of universal basic income, fantasies about AI taking over our jobs so we can dedicate ourselves to leisure, or just being blobs floating around getting spoon-fed, are just that—fantasy.”

Want to learn more?

To find out more about Madeleine Ransom’s AI research, visit her website. You can also see further information about work being done to advance AI literacy at the Digital Transparency Research Excellence Cluster.