"In order to work with AI, we must work on being human"

Artificial intelligence has entered into the decision-making frameworks of many, if not most, industries. While great progress is being made at breakneck speed, one of the key issues that has arisen is the ethics that guide these AI decisions—or the lack thereof. Without the consideration of ethics—whether a decision is good, bad, right or wrong—undesirable consequences could be commonplace.

At just 42, Dr Cansu Canca has carved for herself a unique position as an AI ethicist. Named one of the top thinkers in AI ethics and governance by the World Economic Forum in 2023, she has delivered over 100 talks on the topic at institutions including Harvard Business School and the United States Department of Justice.

She discovered that NUS offered a PhD programme that was hard to resist, even with scholarship offers from universities around the world.  “When I chose Singapore, my mum was like, ‘I’m so glad you’re not going too far’,” she added. “Singapore was much easier for my family when I was a student living there [due to the visa restrictions they would face from nearer countries in Europe]. I loved my education at NUS. It was probably one of the best decisions I made.”

Dr Canca highlighted the “amazing” financial and research support she received from NUS. “I ended up going for research stints at the World Health Organization (WHO) in Geneva and the Harvard School of Public Health in Boston, but all of these were supported by grants from NUS. The philosophy department has really a high-end, world-class faculty,” she added.

RAISING UP ETHICS IN A WORLD OF AI

Dr Canca described philosophy as “an organised approach to the world’s hard questions. [It entails] organising your thoughts, making them structured and systematic, and going in a direction where you can have some answers”.

She emphasised that ethics is fundamental. “How are people treated? Are they suffering? Are they happy? How can we make it better? And finally, how can we make the life experience better? That is the most fundamental question if you care about anything.”

Her doctoral research at WHO focused on organ transplantation, after which she became a lecturer in Applied Ethics at the University of Hong Kong. There, she designed a medical ethics curriculum and taught bioethics.

She is attracted to the urgency of applied ethics. “You have to give answers: there is a patient and there must be a decision about that patient's treatment. For example, who is going to get the kidney? Who is going to get the life support? Who is going to get the vaccination? You don’t have the luxury of thinking about this for centuries; you have to make a decision. You have the moral weight of any action, including inaction,” she described. “I have always been very drawn to the idea of combining this importance and urgency with philosophy’s very structured and deep thinking process, which leads you to robust answers.”

It was while she was at the University of Hong Kong that she noticed how technological tools were being used to make value-based decisions in hospitals—often bluntly. For example, she explained, an AI system predicting which patients need more healthcare might prioritise those who return to the hospital more often. At first glance, this looks objective. But in reality, wealthier people can afford more frequent visits, while lower-income patients might delay care due to the cost. The AI algorithm, if left unchecked, would result in deprioritising patients that need healthcare the most. “So whatever pattern the AI system recognises is really dependent on the data and values encoded in it. If we don’t know that, we miss a whole chunk of the ethical assessment that we make,” Dr Canca said. 

This led to her leaving Hong Kong for Cambridge, Massachusetts, to research and affect how AI makes ethical assessments. “I thought [it would be good to go back] to Harvard,” she recalled of her move in 2017. “This town is full of researchers, and I was sure they were working on [the ethics of AI]. But they were not working on it—not in relation to the practice and to the practitioners.”

She also talked to professors at Harvard and the Massachusetts Institute of Technology (MIT), as well as major tech companies, only to learn there was no opportunity to focus on this specific question. So, Dr Canca started her own consultancy, AI Ethics Lab. She went to a café, bought a website, created a logo and started contacting people. “I said, ‘Can we work together? I have nothing to offer, no money—nothing—but we can start working on these research questions, developing papers and workshops to guide the practitioners. And it worked because there were no similar groups—which meant there was no real competition. So if anybody wanted to work on this, I provided the setup.”

The AI Ethics Lab team consists of philosophers and computer scientists — an unconventional combination that, to Dr Canca, makes perfect sense. She pointed to a precedent in public health ethics: a group of philosophers within a programme called Ethics and Health, who advised on triage decisions during the 2010 Haiti Earthquake. These experts worked together to make extremely difficult ethical decisions that had to be implemented immediately. “So in my mind, that has always been the model,” she explained. “You bring the computer scientists together with the philosophers, and they together make better decisions, because computer scientists know the technical aspects and philosophers know how to determine what is right.”

Her timing was impeccable when she launched AI Ethics Lab in 2017. The following year, the Cambridge Analytica and Facebook scandals, in which personal data belonging to millions of Facebook users was used without consent by Cambridge Analytica to market targeted political messages through AI, heightened awareness of ethical risks in technology. “I was in town, I was available, and I had a very easy-to-find website,” she said. “I was already connected to the academic community here, so I was pulled into different projects, and I pulled in my other collaborators into those different projects.” The lab quickly gained traction, leading to talks, workshops and consultations with companies.

As an early voice in AI ethics, Dr Canca has worked with organisations such as the Department of Health and Human Services and the Department of Justice in the US, educating them on the ethical implications of using AI tools. She also collaborates with the World Economic Forum, the United Nations and INTERPOL to develop guidelines for law enforcement, investors and companies on responsible AI innovation.

Although the world is enamoured with AI now, Dr Canca argues that the key to working with the technology lies in strengthening human qualities. “The reality is, in order to work with AI, create better AI systems and be successful in an AI-embedded world, it’s much more important that we work on being human,” she said. “We need to be more creative, more critical in our thinking and we need to understand the world better. That’s how we can train and control AI better.”

Text by Theresa Tan