ASTANA – This is the second part of The Astana Times interview with Richard Stirling, co-founder and CEO of Oxford Insights, a global consultancy firm behind the Government AI Readiness Index. In the first part, he discussed what it takes for a government to be AI ready and the opportunities he sees in Kazakhstan. In this part, he delves into broader risks stemming from AI.
Flip sides of AI
There are, however, broader risks associated with AI. The downside is that if people rely on just a few AI models—mostly trained in the U.S. or China—they may not reflect daily lives, experiences, or languages.
Richard Stirling, co-founder and CEO of Oxford Insights. Photo credit: oxfordinsights.com
“As a result, the answers that we get back are perfectly valid, but they reflect the data upon which the models are trained,” Stirling said.
He mentioned cases of AI models assisting in court decisions in the U.S.
“It worked pretty well for a while until they tested the model and realized that the decision was purely being taken on the ethnic makeup of the person who was in front of them. That sounds like a terrible AI story, but it had learned that from the decisions of the court before. It is holding up a mirror to our society, and our society is not free from bias,” Stirling said.
Because no society is free from bias, relying solely on existing data without addressing these issues risks reinforcing or even deepening societal biases. While using AI in areas as delicate as court or healthcare might raise concerns, this is what every country needs to decide for themselves, said Stirling.
“I probably wouldn’t have it in court, and I probably would have it in medicine. In courts, you are taking away somebody’s life and liberty, and in medicine, you’re giving it back. But I can completely understand why you wouldn’t be happy with it in medicine. I think that these are the discussions that we should be having as a society, and we should reach a societal conclusion. We should put boundaries in place that reflect that conversation, rather than being led by people who are over-enthusiastic on the technology, or over negative on the technology,” Stirling said.
Data privacy
The other flip side is that rapid AI development and integration raises critical concerns about data privacy and security. Stirling calls for updating the rules and laws to “reflect the challenges that society is facing now.”
“Data privacy is one thing, and making sure that the data that you put into these systems ends up being handled in the way that you want it to be handled. (…) What you could consider to be personal data? Is your image personal data? If I took a screenshot now, should I be able to use that? Should I not? And then there is putting out content that reports to be you but isn’t, so putting out fake content. And what’s the role in that?” he said.
These debates remain unresolved as laws struggle to keep pace with the rapid evolution of technology.
“These are live questions, which I don’t think anybody has properly got to the bottom of. Because the moment you codify something in a law, it becomes difficult to change. What is also true is the hour of technology. It is changing week to week at the moment,” he said.
The European Union, for example, has been proactive in establishing regulations on digital rights and AI, addressing issues such as image usage long before AI-generated content became mainstream. Their laws recognize the right to privacy as a fundamental human right.
In 2024, the EU’s AI Act entered into force, outlining clear guidelines and responsibilities for developers and users of AI in specific applications. “Because the risk with regulating too hard and too early is you stifle an industry you want to grow, and you need to give people the space for creativity,” Stirling said.
A little expert in the pocket
AI can transform not only industries but also become a helpful assistant in the hands of people. The uses may vary from planning a trip itinerary to doing a summary of a thousands-page book. Stirling described AI tools as a “little expert in my pocket.”
“I will confess to being somebody who uses AI a lot. I travel a lot with work. I use AI to work out what’s a good itinerary for me when I turn up in a new city. (…) I take pictures of rooms in my house and ask for suggestions on color, or use AI to provide me with advice on the steps needed, DIY or anything in a regime to get fit,” Stirling added.
