In April 2023, our research team launched the report Training the News: Coverage of Canada’s AI Hype Cycle (2012–2021). The three specific objectives of the report include to:

1. Examine the newsmaking practices and processes through which (tech) journalists try to objectify AI as a matter of everyday concern or a sociopolitical issue;
2. Identify actors, institutions, organizations, and issues that shape discourse on AI in legacy media as a way to chart participation and influence in AI coverage; and to
3. Analyze the formation of AI controversies and their rhetorical closure in legacy Canadian media.

Building on two-year long research project, we present key findings and recommendations.

AI News is Business News

Over the last two years of research on artificial intelligence coverage in Canada, we made two significant findings. First, we found that AI news is business news. And second, we noted that AI coverage is largely populated by machine learning experts. These two, we argue, are related.

These findings result from a multinational and multidisciplinary research project called Shaping AI: Controversies and Closures in Media, Policy, and Research. In the project, four teams from Canada, France, Germany, and UK examine the social construction of AI in their respective countries. In Canada, we paid a particular attention to how legacy media has shaped public discourse on AI. Building on 14 interviews with journalists and a computational analysis based on more than 7,000 articles in French and English mainstream newspapers, we found that AI coverage closely aligns with how businesses and governments frame AI. According to our computational analysis, stories on AI primarily focus on finance, international relations, commerce, economics, lab funding, and potential applications of AI. As an interlocutor suggested, journalists “tend to present emerging technologies in glorious terms. So, 90% of the time, these technologies are featured in a way that is very ‘wow.’” Put differently, legacy media often portray AI uncritically as a resource that can be harnessed to revolutionize our economy. Nothing less.

This isn’t necessarily surprising. Across the world, a number of countries, including Canada, have engaged in a global race where AI is a prized resource that must be appropriated in order to eventually trigger an economic boom. In Canada, federal and provincial governments have set up para-public institutions to meet this very objective. Scale AI or Forum AI Québec act as building blocks for the construction of a political economy of AI, dedicated to the development and deployment of machine learning techniques for local businesses. In legacy media, such a governmental infrastructure have generally received good press, even though tangible results have yet to materialize.

Who is making the news?

To report on AI, journalists typically rely on the expertise of computer scientists. As one journalist put it, “who is the best person to talk about AI other than the one who is actually making it?” This kind of journalistic reflex has translated into a concentration of a chosen few experts in Canadian AI coverage. Our research found that the same computer scientists, and the institutions and corporations that employ them, dominate AI stories. In comparison, social scientists and critical voices struggle to make their way into public discourse on AI.

When interviewed, computer scientists rarely use the opportunity to explain the technical aspects of AI—to inform the public about their field of expertise. Instead, machine learning experts tend to simplify AI for a lay audience, and in the process, dramatize what AI could eventually accomplish. In a nutshell, they promote a particular vision of AI and shape collective expectations about the future of AI may hold.

This was particularly prominent in light of the recent publication of an open letter calling for a ban on AI experiments. When computer scientists Yoshua Bengio, who signed the letter, did a local media tour in the ensuing weeks, he warned us that AI could become a threat to democracy and world order. But in doing so, the 2018 Turing awardee also cast AI research into a Manichean dichotomy: the bad version that “no one…can understand, predict, or reliably control” and the good one—the so-called “responsible AI”.

Foredooming AI as a future threat is a means to close debate on actual harms generated by current generations of AI. But it is also couched as a call for greater investment of resources in the development and deployment of an “ethical AI.” These interventions are as much about shaping visions about the future of AI as it was about hyping up responsible AI.

Key Findings

1. Tech news tends to be techno-optimistic.
2. There are no significant discrepancies in AI coverage between English and French newsrooms.
3. In Canada, AI is business news not science or technology news.
4. Gadgets, self-driving cars, or other applications are more newsworthy to journalists than the social or technical nuances of AI.
5. AI coverage followed a hype cycle.
6. Computer scientists prevail as key experts in AI.
7. There is little to no media scrutiny on AI research funding in Canada.
8. Ethics dominate public discourse on AI in legacy media.
9. News publishers rely on AI, but they do not discuss AI’s implications for journalism.

Recommendations

1. Promote and invest in technology journalism.
2. Avoid treating AI as a prophecy.
3. Follow the money.
4. Diversify your sources.
5. Encourage journalistic collaboration between journalists and newsrooms and data teams.

For more information, download the report here (or ici, for the French version). Appendixes are also available in English or in French. Consult our op-ed, in both English and French, for a brief exposé on the state of AI coverage in the country or watch the launch event for a riveting discussion on the role of legacy media in shaping public discourse on AI.