Updated data privacy laws are key to public trust in AI

Data is the currency of our digital society. We’re collecting, processing, storing, and reusing an ever-growing volume of data, often devoid of context, considerations about where it came from, or how it is being remixed. New ways of using data, such as training artificial intelligence (AI) systems, continue to drive us further down this path. Increasingly, data is seen primarily as fuel for digital innovation, specifically AI. But AI has a problem that more data will not solve: it lacks public trust. A recent study notes that 27 per cent of Canadian companies have banned AI citing legal and privacy concerns as key reasons. The latest Edelman Trust Barometer finds that 54 per cent of Canadians surveyed reject AI innovation, while only 17 per cent embrace it. Anecdotally, I hear similar concerns from clients and other professionals working in the AI governance space. If we want to build a responsible and trustworthy AI ecosystem then we need to take data privacy seriously by strengthening privacy regulations. Perhaps this feels antithetical. Yet, the Edelman report finds that the government's inability to effectively regulate new technologies puts trust in innovation at risk. Privacy is a core element of what needs to be addressed in effectively regulating digital technologies which rely on the use of data. As it stands, Canadians should be concerned. Our current privacy laws are antiquated, and based on upholding the comforting myth that data can be anonymized to remove the parts that would compromise privacy, specifically personally identifying information (PII). A growing body of research shows that anonymization is not possible when we have big data that can be easily recombined to re-identify individuals, a risk that grows with increased datafication. Furthermore, data that might not be considered PII, but that involves human behaviour, can also be used in ways that create risks and cause harm. Our data doesn’t just impact us, it also impacts other people who are algorithmically determined to be similar to us—a lesson we should have learned in the years following Cambridge Analytica. The nature of machine-learning techniques relies on the ability to use aggregated data from individuals to make predictions about groups of people, which troubles the notion of legal protections focused solely on the individual. Instead, legal scholars have argued that privacy is interdependent in nature. Our narrow perspectives about data mask issues that are central to the conversation around how data is actually used in AI systems, and why it's deemed valuable. In addition, consent mechanisms for data use are broken. The notion of giving meaningful, informed consent online is obscured by long and cumbersome terms and conditions concealed by digital ‘clickwrap’. More recently, some organizations are trying to retroactively redefine the terms of consent by posting disclosure notices for new and potentially lucrative uses for data, such as training generative AI. While consent remains important, it may no longer be sufficient in and of itself to provide adequate protections for all people in a digital context. Modernizing Canadian privacy laws involves rethinking our perspectives about data. Data is an extension of people, determining our access to fundamental goods that determine life chances such as education, healthcare, financial resources or social services. Taking a broader human rights-centered approach can address current issues while providing a solid foundation upon which to guide future digital technologies. In her June 2020 paper, A Human Rights-Based Approach to Data Protection in Canada, legal scholar Teresa Scassa has written extensively on this topic. One concrete step in the privacy reform process could be to include a set of recitals that would enshrine privacy as a human right, and “give legislative voice to the principles and human rights values that are meant to underlie data protection law in Canada,” as Scassa writes in the report. Embracing and prioritizing a human rights perspective allows us to seek clarity as to how data involving humans should or should no be used in this digital era. To truly embrace the principles of autonomy and justice, they must extend to our digital selves—our data—as a fundamental entitlement. Though it sounds paradoxical, this isn’t an affront to innovation. It could, instead, be part of sustainable, trustworthy innovation based on the public good. Real trust needs to be earned and freely given when parties act in ways that are trustworthy. Providing comprehensive protection for the privacy interests of Canadians that aligns with human rights, could engender greater public trust and advance more responsible AI development and adoption. Katrina Ingram is the founder and CEO of Ethically Aligned AI, a company focused on helping organizations to drive better outcomes in the design, development and deployment of AI systems. She is the former data ethics adviser for the City of Edmonton, an IAPP-certified information privacy professional, and has been named to the "100 Brilliant Women in AI Ethics" list.Update: This op-ed's headline was changed to "Updated data privacy laws are key to public trust in AI" on May 6. The Hill Times

Updated data privacy laws are key to public trust in AI

If we want to build a responsible and trustworthy AI ecosystem then we need to take data privacy seriously by strengthening privacy regulations.

Time for a shift: Canada must treat Alzheimer’s like other progressive diseases

Canada faces a cybersecurity crisis with critical infrastructure at risk

Let Ukraine strike back

Who’s standing up to the danger of World War III?

The sad saga of Afghan-Canadian military advisers

Between wars and fears of fascism in the U.S., there’s plenty to be scared about

Bill C-27 lacking transparency for private-sector, AI privacy intrusions

Canadians demand privacy, transparency, and personal data control to build trust

‘The devil’s in the details, and we don’t have any’: critics, civil liberties groups decry feds’ lack of clarity on changes to privacy and AI bill

Necessary but not enough: Canada’s proposed new private-sector privacy law

The ground rules for the industrial adoption of AI

News

Get free news updates

Opinion

Submit a guest opinion article

resources

Advertise

subscribe

get help

Updated data privacy laws are key to public trust in AI

If we want to build a responsible and trustworthy AI ecosystem then we need to take data privacy seriously by strengthening privacy regulations.

Get The Policy Briefings Newsletter