00:00:00:00 - 00:00:22:03
Grant McDonald
AI is everywhere. ChatGPT and other generative tools have been around for a while. But if your feed looks like mine, you're scrolling past AI headlines and maybe even AI generated content every day. I'm your host, Grant McDonald, and on the Banking on AI podcast today, we chat with someone who spends a lot of time thinking about what that means for journalism.
00:00:22:05 - 00:00:51:12
Grant McDonald
Natalie Turvey is the Executive Director of the Canadian Journalism Foundation, which champions excellence in Canadian journalism through fellowships, talks and awards and works to strengthen the understanding between the media and the organizations it covers. TD proudly supports CJF as a platinum sponsor. Recently, CJF launched the Hinton Award for Excellence in AI Safety Reporting, recognizing evidence-based journalism that explores AI risks and solutions for responsible innovation.
00:00:51:14 - 00:01:15:16
Grant McDonald
At TD, we share that commitment. Our human-centered, Trustworthy AI approach guides every AI project at the bank. Today, Natalie joins us to talk about journalism's role in shaping responsible AI and how this fast-moving technology is transforming her industry.
00:01:15:18 - 00:01:36:15
Grant McDonald
Natalie, thank you so much for being here. We really, really appreciate it. I'm really intrigued about the conversation we're about to have, you know, our worlds might seem far apart. Banking. You represent the media industry, but I'm very curious as to the connections that you see. And then on top of that, how you're seeing AI working right now across journalism.
00:01:36:17 - 00:02:02:22
Natalie Turvey
Well, it's a pleasure to be here. And thank you for the opportunity. We're seeing AI being used by newsrooms largely as an assistant and not a replacement. It can help save time, complete tedious tasks, analyze a large amount of information. But always with human oversight and robust AI editorial policies in place by newsrooms that are communicated to the Canadian public.
00:02:02:17 - 00:02:12:22
Grant McDonald
So in the work that you do, you're bridging communities. So my world, your world, you are sometimes the connective tissue. I'd love to hear a little bit more about that and sort of the mission that you're on.
00:02:13:00 - 00:02:59:17
Natalie Turvey
Absolutely. So the CJF's mission is to foster excellence in journalism. We do that by supporting journalists and their organizations at every stage and across the country through our programs, things like our monthly talks, our fellowships, training, research and mentorship. And the foundation was created by Dr. Eric Jackman 35 years ago, a Canadian business leader, to help bridge the gap between the media and business communities and our forums and events bring together business leaders and media leaders for meaningful dialogue and this helps build trust, clear up misunderstanding, understand the challenges that both sectors face, and ultimately help support more informed journalism for Canadians.
00:02:59:17 - 00:03:15:02
Grant McDonald
And Canadians depend on the news so much. The work that we do is translated through the news world. When I look at how AI is transforming that, I'd be curious as to what you're seeing in the big newsrooms, but also the small newsrooms that are often not talked about as much but are just as important.
00:03:15:03 - 00:03:39:12
Natalie Turvey
So AI can be a real game changer for the smaller newsrooms. They're already stretched thin, and it allows them to do more with what they have. AI can supercharge investigations so a single journalist can look at municipal documents or other local filings, and quickly find patterns and anomalies to support their reporting. And that might have taken a full team to do.
00:03:39:14 - 00:04:02:01
Natalie Turvey
Smaller newsrooms are also using AI to repurpose content to reach more audiences so they can take a strong piece of local journalism and create other formats, like newsletters, summaries, social media copy, or even scripts for video. And it allows them to reach audiences across platforms that might have required separate teams, to reach those.
00:04:02:03 - 00:04:18:14
Grant McDonald
So they're able to work all these different areas where before it would, just they simply wouldn't have the staff to do it. You touched on high level. I'd love to know if you have examples of what you've seen as well, just in terms of the transformation of AI and how it's helping. I guess journalism as a whole, not necessarily just to small newsrooms.
00:04:18:16 - 00:05:25:10
Natalie Turvey
AI is being used by newsrooms as an assistant and not as a replacement. It helps save time. It speeds up workflows. It can complete tedious tasks and process vast amount of information, but always with robust AI editorial policies in place and human oversight to ensure that information is trustworthy and accurate for Canadians. So a few examples we're seeing to speed up workflows are tools like Otter and Trint that can take a long form interview and transcribe that in a matter of minutes. This work would have taken hours for a single journalist. It can also provide insights into audience engagement with specific stories, and it helps news leaders make better decisions about the stories they cover in their newsrooms. And it can process vast amounts of data, like court filings, public spending.
And that would have required a full data team before these tools were accessible to newsrooms. So the theme is AI is assistant, not author in producing quality journalism.
00:05:25:15 - 00:05:47:23
Grant McDonald
And you know, I remember a number of years ago when some tools were being introduced, there was a little bit of hesitancy towards it. I remember a moment changed where someone using this piece of technology, talked about it, how great it was, and then another colleague didn't buy in at all. And now it's at the point where this colleagues who didn't want to ever use it, will not go to work unless they have access to these tools.
00:05:48:03 - 00:05:50:22
Grant McDonald
Are you seeing that mind shift that, that change as well.
00:05:51:00 - 00:06:06:18
Natalie Turvey
That we have to keep in mind these technologies are still nascent. They're prone to making mistakes, fabricating information or hallucinations. And the journalists I talked to don't yet trust the outputs without the proper verification.
00:06:06:19 - 00:06:20:06
Grant McDonald
And that's, that's a great lead into what I want to talk about next, which is misinformation. And some of the work that you're doing in that world. What does that look like for you? What, what are you trying to push forward, specifically for journalists, but then the public as a whole?
00:06:20:08 - 00:06:49:07
Natalie Turvey
So this is one of the biggest challenges we face. CJF's own research shows that 88% of Canadians are concerned about AI-generated misinformation, and almost half encounter it in their feeds daily. So the risks are two-fold for journalists. Deepfakes are cheap and they're convincing, and they're everywhere. And this is leading to a public erosion in trust. It's hard to know what is trustworthy information and what to believe.
00:06:49:09 - 00:07:17:07
Natalie Turvey
The risks for the public are even greater because seeing is no longer believing. These deepfakes and AI-generated images are so convincing. And this is why it's important that news literacy skills become an essential life skill. That before the public shares a piece of information, believes that information that they spend some time verifying the source, that this came from a credible newsroom.
00:07:17:11 - 00:07:31:16
Natalie Turvey
Was it covered in other outlets? Does this provoke an emotion or seem too good to be true? And Canadians are taking responsibility for their own literacy in this new AI and digital deception environment.
00:07:31:18 - 00:07:53:04
Grant McDonald
And you know, it's so interesting that you, you point out the idea of on your feed, what you're seeing. And a lot of people wouldn't even realize that it's misinformation. They would just assume and go forward. So it's a great point you're talking about in terms of do that extra step to verify. Some of the work that you do, as well as giving out lovely awards, wonderful awards for the work that is being done.
00:07:53:06 - 00:08:01:14
Grant McDonald
I'd love for you to talk a little bit about the Hinton Award. This is a new one, and I'm curious as to why you brought this forward and why now?
00:08:01:16 - 00:08:41:08
Natalie Turvey
So we're incredibly proud of this one. It just launched, the CJF Award for Excellence in AI Safety Reporting. The award was created with the AI Safety Foundation, and it honours and recognizes journalism that digs deep into the catastrophic risks and misuse of this technology. Things like cyber threats and autonomous weapons. And it was named, after Dr. Geoffrey Hinton, AI pioneer called the godfather of AI, but also one of the strongest public voices about responsible use, development and transparency around this technology.
00:08:41:10 - 00:08:53:20
Natalie Turvey
And journalists play a vital role in informing the public about the potential risks of AI. And this is becoming vital. So we are working to elevate and support that work.
00:08:54:01 - 00:09:11:18
Grant McDonald
That is amazing. I mean, for us at TD Bank, you know, fraud, cybersecurity, these are all top, top, top priorities. And having the relationship with organizations like yourself and journalists to help tell these stories of ensuring the public is staying safe, like I think that is a huge bridge between the two organizations.
00:09:11:20 - 00:09:46:21
Natalie Turvey
And there's a huge sense about how culturally pervasive this technology has become. TIME magazine just honored, the architects of AI as its person of the year, much like they honored the computer in the 80s. And this is why timely journalism is so vital. There has been a sense that a lot of the development has been done with an ask for forgiveness rather than permission approach, and why the CJF feels the time is now to draw attention to some of the risks, the perils, not just the promise of this technology.
00:09:46:22 - 00:09:53:04
Grant McDonald
Absolutely. And you're looking at this constantly. I'd love to pull out some trends that you might be seeing.
00:09:53:06 - 00:11:01:06
Natalie Turvey
So there are a few big ones. One is search generative experiences. And these are the new AI summaries or full answers that you're getting in your search queries. So instead of driving the audience to news websites, you're getting that full answer. And while it's convenient for the searcher or the audience, it's taking away traffic from news organizations.
And this could have an effect on financial sustainability. So to survive search generative experiences, it's vital that newsrooms bring exclusive, quality journalism to their audience to drive subscriptions, and they optimize their content in this new world of AI summaries. Another big trend that I'm sure everyone's following are AI and copyright battles. Many of the AI, the LLM models have been trained by scraping the web without proper compensation for news organizations or the creators of that content and how that gets resolved, whether it's through licensing agreements or other payments, could have, financial implications for newsrooms.
00:11:01:07 - 00:11:14:16
Grant McDonald
And that's it. You know, it's, it's original content, it's journalism. It's, you know, boots on the ground finding the story versus scraping the web. I'd be curious to know if there's a couple stories that kind of stand out to you where you're saying, that's really good reporting on AI
00:11:14:17 - 00:11:39:09
Natalie Turvey
I've seen a few examples. I always find the profiles with the AI insiders like Geoffrey Hinton and Yoshua Bengio, resonate with the public. When the creators of this technology are raising alarm bells about its risk, people pay attention and it cuts through the hype of the valuation of these big companies and brings home to the public the ethical stakes.
00:11:39:09 - 00:13:17:23
Natalie Turvey
So those are important. A story that resonated with our journalism community was a recent one that appeared in The Local by Nicholas Hune Brown, called "A Possible Scammer in Journalism's AI Era", and that reporting exposed what appeared to be a completely fabricated freelancer. So a person who sent in a pitch was published in internationally recognized outlets like The Guardian.
And this reporting exposed that many of the sources and quotes in those articles were fabricated, and the writing was produced by AI. And some of the public good was these articles, of course, were taken down. But the newsrooms that I speak to are instituting protocols now to vet their freelancers. And one might be talking to the previous assignment editor or ensuring there's a policy for a video chat with that freelancer so you can look them in the eyes and talk about the story and verify they're a real person.
On the positive side of AI, I came across a human interest story in the New York Times saying, "doctors told him he was going to die. AI saved his life". And it was a patient who had a rare blood disorder, conventional drugs stopped working. So through an AI a model specializing in drug repurposing, it was able to scan vast amounts of data and information and find a complex cocktail of drugs that humans hadn't thought of trying.
The doctors administered the drugs, the patient went into remission, and it saved his life. So that's an example of how AI can be a super partner when humans hit a wall.
00:13:18:01 - 00:13:28:16
Grant McDonald
Now, you talked a little bit about smaller newsrooms, some larger newsrooms, but I'd love to hear some examples of the investigative side, just that really, really meaty journalism and, and how AI is kind of part of that.
00:13:28:18 - 00:13:57:02
Natalie Turvey
There's some exciting work being done in investigative journalism assisted by AI. Canada's National Observer built an AI-powered tool called Civic Searchlight. It allowed them to comb through more than 550 municipal meetings. And through that work, they uncovered an AI-created targeted misinformation campaign, and their reporting had real public impact. It led to an Alberta town staying in a climate protection program.
00:13:57:07 - 00:14:24:19
Natalie Turvey
Another great example comes out of CBC. They did a story called "Investigating Doctor TikTok". And for a story on health misinformation on TikTok, they used an AI tool to isolate relevant video making health claims. They went through thousands of videos on the platform. Then their journalists were able to find patterns and contextualize the results for Canadian audiences so they could protect themselves against health misinformation.
00:14:24:21 - 00:14:43:07
Grant McDonald
And that work being done just it could have been done, but it would maybe take years. That is the difference between the tools being used. And to your last point there, in terms of Canadians protecting themselves, I love the bridge because on the day to day, if you're reading a long investigative form from the CBC, you can trust that it's, you know, it's good.
00:14:43:09 - 00:14:56:10
Grant McDonald
The research is there. For your everyday scroller, how are Canadians supposed to protect themselves from misinformation? Disinformation? Just completely non information that looks legitimate. Are there any tools or any resources?
00:14:56:12 - 00:15:22:15
Natalie Turvey
From our research, we've determined Canadians know they have a responsibility in protecting themselves against misinformation. And a lot of the tenets of news literacy can be applied to information literacy. So that initial pause before sharing an article, verifying the source, you can apply that to information you've come across in your industry, whether it's phishing or smishing scams.
00:15:22:17 - 00:15:50:20
Natalie Turvey
Credible looking video featuring political leaders or athletes around investment schemes, even, what we're seeing with AI-generated voice that can replicate a family member in the grandparent schemes, protecting yourselves even through things like, having a safe word in your family, to ensure that you're verifying a real person can really, really help Canadians protect themselves in this environment.
00:15:50:22 - 00:16:24:05
Natalie Turvey
In terms of tools and skills, the CJF does a lot of work around news literacy, helping the public understand the value of fact-based journalism. Craig Silverman, a big misinformation thinker, has developed, a platform called Indicator. And, this is bringing tools and skills to journalists, researchers and engaged citizens, not only highlighting targeted AI misinformation campaigns, but also giving them the skills, the verification skills, to protect themselves in this environment.
00:16:24:08 - 00:16:41:03
Grant McDonald
So there are some tools out there and just you got to think for yourself as you're going through that information. You know, there's, there's the scary side. There's the very interesting and positive side. I always like to wrap with an understanding as to what gets you excited about the world of AI ahead of us in the world, that you kind of work within.
00:16:41:05 - 00:17:15:01
Natalie Turvey
What gets me most excited is how AI can support smaller or local newsrooms. This is such a vital part of the news ecosystem, bringing Canadians the information they need about their communities. And we're seeing AI as a huge support to bring those Canadians the stories that matter. The CJF has just launched a year-long program called the CJF Next Gen News Creator program bringing some of these AI skills to journalists to help them better optimize their content and reach more Canadians.
00:17:15:01 - 00:17:17:14
Grant McDonald
That is very exciting indeed. Natalie, thank you so much for joining us.
00:17:17:17 - 00:17:18:10
Natalie Turvey
My pleasure.
Want to learn more about TD podcast?