Transcript: How the Canadian book industry is approaching AI
Go back to the main session page.
Nataly Alarcón: Hello, everyone. Welcome to Tech Forum, an educational series brought to you by BookNet Canada. I am Nataly Alarcón, BookNet’s marketing and events manager. Welcome to How the Canadian Book Industry is Approaching AI.
Before we get it started, BookNet Canada acknowledges that its operations are remote, and our colleagues contribute their work from the traditional territories of the Mississaugas of the Credit, the Anishnawbe, the Haudenosaunee, the Wyandot, the Mi’kmaq, the Ojibwa of Fort William First Nation, the Three Fires Confederacy of First Nations, which includes the Ojibwa, the Odawa and the Potawatomie, the Métis, as well as the unceded and ancestral territory of the Musqueam, Squamish, and the Tsleil-Waututh peoples, the original nations and peoples of the lands we now call Beeton, Guelph, Halifax, Thunder Bay, Toronto, Vancouver, Vaughan, and Windsor. We encourage you to visit the native-land.ca website to learn more about the peoples whose lands you are listening from today.
Moreover, BookNet endorses the calls to action from the Truth and Reconciliation Commission of Canada, and supports an ongoing shift from gatekeeping to space-making in the book industry.
So, here’s the agenda for today. First, I will share background information on the survey and how we got here. Next, I’ll share insights into the survey results from Canadian respondents, first, from the individual perspective, and then we’ll take a look at the organizational perspective. Then we’ll take a look at some of the concerns and pain points respondents have, followed by high-level takeaways. And next, we’ll share links to AI-related resources that go beyond the survey. And to wrap things up, my colleague Lauren Stewart will join me for the Q&A portion of the webinar. So, please use the Q&A panel found in the bottom menu to submit your questions throughout the presentation.
Also, the slides and the recording from this session will be available in the Tech Forum Content Library website, in a few days, and we will share the link via email.
Okay. So, let’s talk about the survey and how we got here. Back in 2023, BISG created the AI Working Group. This group was created in response to the growing interest in, and uncertainty around, the use of AI across the book supply chain. Together, the Working Group members, which includes representatives from BookNet, developed the first iteration of the survey, which was fielded in 2025. This survey received 559 responses, 100 of them from Canadians. In 2026, the survey received a total of 771 responses, 284 of them from Canadians. That is a 181% increase in survey participation, which we were very excited to see.
And what did we do with all the data that we collected? We shared with you, the public. The AI used across the North American Book Industry 2026 report is available on our website and the BISG website for free. My colleagues will drop the link in the chat.
In addition to the report, both organizations are using the data to inform the programming of webinars like this one and also like our mini-series that we had recently, Get Discovered. And we’ll also share a link in the chat. In addition to this, we are also sharing sector-specific information with organizations such as library groups when available.
But why do this? I mean, yes, the BookNet team loves a good survey, as probably all of you know. But why AI and why now?
First of all, we want to empower the industry and people like you who work in it with high-quality and easily accessible data that is relevant to your work and the market that you are a part of. We also want to identify needs for guidance, best practices, and the standards development or improvement to act on those needs or support relevant initiatives. And finally, we want to keep a finger on the polls as the industry and the technology, including the legislation created around it changes.
And today, I will be sharing insights from the survey from Canadian respondents only. So, this means that the insights you will see here won’t necessarily match the data from the report, which takes a North American English speaking market approach as a whole. Though, spoiler alert, there is alignment between the U.S. and Canada in key areas.
So, first, let’s start by learning about the Canadians that took the survey. One quick note before I continue. You may notice that I will verbally share the majority of the slides’ content or key information, and I’m doing this intentionally for accessibility purposes.
So, the first thing we learn is that 65% of respondents work for an organization, and 35% of them are self-employed or independents. Let’s take a look at where those who belong to an organization work.
From the group that said that they belong to an organization, 41% work for a publisher, 29% work for a library, 9% work for an independent bookseller, and every other group accounted for 6% or less in terms of representation in the data. From the group that works for publishers, we learned that 77% of them work for a trade publisher, 18% work for a university press, and everyone else accounted for 3% or less. In terms of the departments, in case you’re curious, editorial and sales were the most represented in the data, followed by production and marketing.
We also asked self-employed respondents to tell us about their line of work. Twenty-four percent are self-publishers or author publishers, 21% are editors, and 18% are authors working with trade publishers. In addition to this, 19% of self-employed respondents chose Other to indicate their line of work was not listed. Every other group, consultant, indexer, among others, accounted for 7% or less.
So, what else did we learn about Canadian respondents? Canadian respondents, as a whole, were almost evenly split between folks who have worked in the industry for 10 years or less at 51%, and those who have worked in the industry for 11 years or more at 49%. Sixty-eight percent of them work for an organization that has been in business for 36 years or more. The next most represented group was those who work at companies that have been in business for 15 to 25 years at 13%. And in terms of employee count, companies with 100 or more employees accounted for 31% of responses. Companies with 21 to 50 employees were the second most represented in the data at 23%. And another group with a strong representation was companies with 5 to 10 employees at 20%.
Okay. With all that context in mind, let’s take a look at the actual survey results. So, how are individuals approaching AI in their work? Right off the bat, we found that, overall, 38% of Canadian respondents at the time of taking the survey were using AI for work, while 62% said that they were not. But how does this compare to 2025, you ask?
When we look at the 2025 and 2026 data side by side, we see that Canadians who said they were using AI decreased 8% year over year. As a side note, if you’ve read the report, you may notice that even though this graph is based on until today and released data from Canadian respondents only, it looks quite familiar. And that is correct. The percentages are the same as for the whole North American English speaking market, which indicates that for this crucial data point, there is a strong alignment between the U.S. and Canada.
Now, remember how we asked all these questions about years working in the industry, organization size, and so on? Well, those insights allow us to slice and dice the data in ways that help us go even further in the analysis. Here in this graph, for example, you can see that professionals who had been working in the industry for six years or less, as well as those who had been working in the industry for 36 years or more, were closely aligned in their responses, with about 30% of respondents saying they used AI for work.
For respondents in this 7 to 10 years bracket, 37% said yes to using AI. For the 11 to 14-year bracket, 43% said they used AI as well. And next is the group with the highest adoption of AI, the 15 to 25 years in the industry bracket, where 47% of respondents said that they’ve used AI. For the 26 to 35-year bracket, 45% said the same, that they were using AI. So, there’s almost a trend that indicates that the longer, but not too long, someone has been working in the industry, the more likely they are to use AI tools in some capacity.
Let’s take a look at use of AI by individuals from another perspective.
So, let’s look at employees here for a second. When we look at individual use of AI for work in relation to the size of the organization they work for, we see that respondents working at the smaller companies were less likely to use AI for work. One second.
The majority of respondents working at organizations with 2 to 4 employees said they were not using AI at 63%. And the same is true for those working at organizations with 5 to 10 employees, 72% of them said they were not using AI. And the data leans to the opposite side for the next brackets. Eleven to 20 employees, 54% of respondents said they used AI. For 21 to 50 employees, 57% said the same. For 51 to 100 employees, 60% agreed with the same. And for organizations with the largest employee count, 100 employees or more, 45% of respondents said they used AI.
So, a reminder that this data is from the perspective of the individual. We will take a look at organizational use later in the presentation.
Overall, based on the data that I’ve shared, looks like employees working for medium to large organizations, but again, not too large, are more likely to, as individuals, use AI for work.
Let’s now take a look at how respondents said they were using AI. So, the types of accounts they used and the areas where the tools were being put to work.
We found that among Canadian AI users, 44% of them used an enterprise or team account managed by the organization they work for. Nineteen percent used a personal account with privacy protections. Eighteen percent used a personal account without privacy protections. This can include, for example, a free account where the user hadn’t opted out of data training. Twelve percent used a combination of enterprise and personal accounts. Four percent were not sure what type of accounts they used. And 3% said other.
Digging a little bit deeper into the data shared in that previous slide, we found that when it came to personal accounts, the group that reported the highest use of accounts with privacy protections was individuals who had been working in the industry for 15 to 25 years. On the other hand, the group that reported the highest use of personal accounts without privacy protections was individuals who had been working in the industry for 7 to 10 years.
So, in conclusion and according to this data, the years an employee has been working in the industry are not necessarily an indicator of their understanding of the types of accounts/agreements, and the risks that come with them that they have available when choosing an AI tool. So, there is a learning and training opportunity here.
Now, let’s take a look at areas where AI was being used. The list is long. So, be prepared. But I believe that it’s important to review it from top to bottom to share a more complete picture of where adoption is actually taking place. So, bear with me.
In first place, we have research at 43%. Some examples of potential uses include market research, title research, and fact checking. This option was new in 2026. So, we don’t have year-over-year data. Next is marketing at 37%. Some examples include copywriting, campaign planning, audience targeting, content creation, and brainstorming. This area saw a 1% year-over-year increase. The administrative area ranked third at 36%. Some examples of applications include scheduling, meeting notes, email management, and slides creation. This area saw the highest year-over-year increase among all areas at 8%. Editorial was fourth on the list at 25%, with a year-over-year increase of 6%. Next is data analysis, which includes the creation of Excel scripts using it for SQL and similar tasks at 24%, with a 1% increase compared to last year. Next, we have metadata and title optimization at 22%, with a two year-over-year increase. This includes tasks related to keywords, BISAC and Thema selection, and Thema code selection, ONIX files management or creation, among other tasks.
Next, we have education and professional development, another new option for 2026 at 15%. Publicity is eighth on the list at 14%, with a 2% decrease compared to 2025. Examples of tasks can include media outreach, press materials, influencer engagement, among others. Next is sales forecasting and market analysis at 13%. This area saw a 3% increase year-over-year. Accessibility, another new option for 2026, took the 10th spot on the list at 11%. One common use of AI in this area is alt text creation. Production, which includes layout design, typesetting, and file preparation, is 11th on the list at 10%, and this area had a 1% increase compared to 2025. Next, we have customer service or reader engagement at 7%, with an 8% year-over-year decrease. So, this is the highest decrease across all areas.
We are reaching almost the end. Code development or management for digital products was 13th on the list at 7% and saw a 2% decrease compared to 2025. And 5% or fewer respondents indicated the following use cases. Translation, manuscript evaluation, rights and licensing management, creating AI-voiced audiobooks, and QA testing.
Now, from this list, there was one particular area we were interested in diving deeper, and that is metadata. So, we learned that from the group of respondents using AI for metadata tasks, 62% of them used it as an assistant to identify potential additions or changes to an originally human-created metadata record. This use case was more popular among Canadians versus U.S.-based respondents, 62% versus 52%. We also learned that 14% were using AI only to help with subject code classification. Here, Canadian respondents were again more likely to do this when compared to their U.S. counterparts, 14% versus 6%. Five percent were using AI to create all the metadata records of front-list titles. This was almost the same for both markets, 6% for the U.S. and 5% for Canada. An additional 5% were using AI to optimise the metadata records of backlist titles. This use case was more popular among U.S. respondents, 5% versus 18% in the U.S.
Okay. So, we’ve learned that 38% of Canadian respondents said they were using AI for work. Now, did the use of AI impact efficiency and productivity? Let’s find out. Seventy percent of Canadians who used AI for work said AI has increased their efficiency or productivity. Here are some comments respondents left sharing more insight into how this was accomplished. The first comment is from a person who works at a library. They said, “I use AI primarily to assist…” Sorry. One second. Give me one second.
“I use AI primarily to assist with research and writing, specifically editing, brainstorming, and wordsmithing my writing for reports. I will say it streamlines the research process, and makes my writing and editing quicker because I find myself getting stuck less.” Here, we have a comment from a professional publisher employee. “AI has helped me address issues in copy, for example, in having a book description meet our guidelines, and pointed me to sources to support statements in books when the original source was an AI hallucination.”
On the flip side are those who said they were using AI, but it had not made them more efficient or productive at 30%. Here are some of the comments from this group of respondents. This one is from an individual working for a trade publisher in the marketing department. “I have tried to use ChatGPT to suggest comparable titles, and have found that their responses produced were totally unhelpful. The amount of time spent with additional prompts to produce more helpful results was greater than if I just did the research myself.”
This person working for a library shared the following. “I have used AI to create apps scripts which replace a few manual tasks such as copy/pasting information between the spreadsheets. But to be honest, these were not usually time-consuming tasks. So, I can’t say that it has had a significant impact. Anything important still has to be done by humans since AI is so unreliable and prone to hallucinating.”
When asked about training and professional development, this is what respondents had to say. Thirty-seven percent of them reported ethical objections to AI training, and that it wasn’t relevant to their work anyway. Twenty-three percent reported ethical objections to AI training even though it is relevant or required in their work. Thirteen percent said that it was relevant to their current role, but that they hadn’t made it a priority yet. Twelve percent said that it was relevant to their current role, and that they actively looked for it. Nine percent said that AI training was not relevant to their current role, but expected that to change. And finally, 5% said that AI training was not relevant to their current role, and they did not anticipate this to change.
Okay. Let’s shift gears and take a look at how organizations were approaching AI at the time of fielding the survey. To get to this, we asked respondents who say they worked for an organization about how their employers were approaching AI. And here is what we learned.
Overall, we learned that over half of organizations were using AI at 59%, 35% of respondents or organizations were not using AI, and the remaining 6% of respondents said they were not sure. In the U.S. market, organizational use was slightly higher at 66%.
But back to Canada. How do these stats compare to 2025? Usage of AI tools among organizations increased 14% year over year. The group of respondents saying that the organizations they work for don’t use AI decreased by 6%. But we noticed that there’s more certainty about whether an organization has adopted AI or not. The group of respondents who say they were not sure decreased by 8% compared to 2025.
This increase in usage is not very surprising if you see it in relation to data from the Canadian survey on business conditions from the Government of Canada. In this survey, they learned that the percentage of businesses that plan to use AI in the third quarter of 2025 increased by 10% compared to the third quarter of 2026, going from 15% to 25%.
But what happens when we take organization size into account? Looking at organizational use of AI in relation to organization size, we found that larger organizations were more likely to use AI compared to those with smaller operations. Ninety percent of employees working for organizations with over 100 employees at their companies used AI. Those who work at organizations with 51 to 100 employees were also more likely to have used AI compared to those with smaller operations at 60%.
On the other hand, of respondents working for an organization with 2 to 4 employees, 57% said their organizations was not using AI, 36% said yes, and 7% said they were not sure. Organizations with 5 to 10 employees had very similar data. And interestingly, in the next group, companies with 11 to 20 employees, that group was evenly split at 50% for each option.
Looking at organizational use of AI in relation to how long a company has been in business, we found that organizations in business between 10 to 25 years reported the highest rate of adoption of AI. Eighty-nine percent of respondents working for an organization in business for 11 to 14 years said that their companies used AI. The same was true for 81% of respondents who work at organizations that had been in business for 15 to 25 years. And the lowest adoption rate was seen among companies that have been in business for 7 to 10 years, where 60% of respondents said their companies didn’t use AI. Everyone else, including newer companies of 3 years or less in business and older companies of 36 years or more, had an AI adoption rate between 50% to 60%.
So, according to the data that we collected in this survey, all this data indicates that the years an organization has been in business do not necessarily impact adoption of AI as much as organization size.
Let’s now take a look at the types of accounts and areas where AI was being used at the time of fielding the survey. The types of tools that were used the most at organizations are AI features within existing software at 55%, which saw a significant increase of 19% year over year. The second most popular was enterprise versions of models such as ChatGPT, Gemini, and Claude at 47%, with a year-over-year increase of 22%. And the third most popular was standard or sometimes free versions of models such as ChatGPT, Gemini, and Claude at 41%. This option saw a small increase of 5% compared to 2025.
Now, let’s take a look at the areas where AI use at the organizational level was reported. In first place, we have administrative at 48%, with a year-over-year growth of 20%. Next is marketing at 47%, with an 11% increase compared to last year. Research ranked third on the list at 33%. Then we have data analysis at 28%, with a 5% change compared to last year. And in fifth place is editorial at 27%, with a 9% increase compared to last year. Next, we have educational at 27%, followed by publicity at 25%, with a 9% year-over-year increase.
Next, we have metadata and title optimization at 25%, with a 5% increase compared to last year. Then we have accessibility at 22%. And in 10th place, customer service or reader engagement at 16%, with a 1% increase compared to 2025. Production at 15%, with a 7% increase. Sales and forecasting at 13%. This area saw an increase of 2% compared to 2025. And next is co-development or management for digital products at 10%, with a 1% increase. Five percent or fewer respondents indicated the following use cases at organizations. Manuscript evaluation, rights and licensing management, translation, QA testing, and AI-voiced audiobook creation. So, very similar to what we saw earlier in the presentation.
So far, in this section, we have learned about the rate of adoption of AI and areas of use. Now, I will share what we learned about how organizations are tackling AI beyond adoption. We asked…not tackling, approaching AI. We asked respondents to choose the answer that best describes the organization’s approach to the use of AI. Here are the top three choices. My organization discourages the use of AI, 25%. My organization is in an exploratory phase, figuring out a definitive approach to AI, 23%. My organization encourages the use of AI within controlled workflows and for specific tasks or departments.
And this is how the Canadian response is compared to the U.S. Canadian companies were more likely to discourage the use of AI, while U.S.-based organizations were slightly more likely to encourage use of AI within controlled workflows. When asked about where respondents perceived leadership, their immediate manager and colleagues were in relation to the adoption of AI, this is what they said. Sixty-seven percent of them said leadership is either pushing for adoption or interested in using AI or adopting AI. Sixty-two percent said the same about their immediate manager, and 48% said the same about their colleagues as a department. But know that their perceived interest in pushing for adoption is much lower for this bracket compared to leadership, 3% versus 26%.
Here are some additional observations. The push for adopting AI among leadership was highest for companies with over 100 employees. And the companies where leadership was least interested in AI were those with 5 to 10 employees. The push for adopting AI from immediate managers was highest among companies with 51 to 100 employees. And the companies where immediate managers were least interested in AI are, again, those with 5 to 10 employees. The push for adopting AI from colleagues was highest among organizations with 11 to 20 employees. And respondents working for small organizations with two to four employees indicated the lowest interest in adopting AI from their colleagues.
So, as we’ve learned, there is no consensus on the interest in adopting AI at organizations. But one thing that I hope we can all agree on is that, now more than ever, it’s essential to have policies or at least guidelines related to AI. So, let’s take a look at what we learned about corporate governance, so, policies, guidelines and such, and what’s happening in that space.
Forty-five percent of Canadian respondents said their company had an AI policy or set of guidelines. Thirty percent said they didn’t have this yet, but they are working on it. And 20% said they didn’t have a policy or guideline, period. The remaining 5% were not sure. The good news is that in the year between fieldings, the number of companies without a policy decreased by 15%. So, this is a really good sign.
And here is how this plays out when we look at the same question in relation to organization size. Respondents from companies with over 100 employees were the most likely to report having an AI policy or guidelines in place at 63%, followed by those from organizations with 2 to 4 employees at 50%. Companies with 5 to 10 and 51 to 100 employees were equally likely to report that they were developing an AI policy or guidelines. And among the small and medium-sized organization, 23% to 29% reported having no AI policy or guidelines, which actually brings me to the next topic, needs for guidance and best practices.
Overall, from all respondents, 71% of them said their organizations would find best practices or guidelines around AI helpful. These are the areas where according to the respondents working for publishers, guidance is needed the most. They could choose multiple selections. So, that’s why the percentages are high. The first one is AI policy development at 62%. The second is laws and regulations, for example, transparency requirements, and copyright at 60%. Next, we have disclosure of the use of AI tools in the creation of books, including the differentiation of AI-generated versus AI-assisted works at 57%. And then we have metadata creation and optimization to aid with book discovery at 50%. This is separate from needs for guidance around metadata creation and optimization to communicate whether AI was using the creation of a title, which was seventh on the list at 38%. And finally, in fifth place, we have ethics at 43%.
These are the areas where according to survey respondents working at libraries, guidance is needed the most. The first is AI policy development at 74%. Next, collections management, 74%. Digital literacy and community education, 72%. Data privacy and security, 67%. And laws and regulations, 67%.
And here is what we learned about the concerns and pain points respondents had with the book industry’s current use of AI. This question was posed to all respondents, and they also could choose multiple options. So, let’s start with the risk-related concerns.
Number one was copyright at 86%. This includes inadequate controls around the use of copyrighted material and certainty about whether AI-generated material can be copyrighted or not, among others. As you can see in the data in brackets, this concern was equally important to U.S. respondents, also at 86%. Next was job loss or negative impacts on creators, so, authors, illustrators, etc., at 81%, followed by concerns around the negative impacts on the environment and/or the organization’s sustainability goals or reporting at 80%.
Now, let’s take a look at the trust-related concerns. The first is AI-generated books, including fraudulent or low-quality content flooding major retail platforms. This was top of mind for Canadian respondents at 91%. Next were concerns about accuracy of AI information when using the tools at 86%. Examples may include false positives from AI detectors and AI hallucinations. The third trust-related concern was related to the accuracy of AI for consumers at 86%. This refers to the inability to distinguish between AI-generated, AI-assisted, and human-generated works, as well as to inaccurate book information generated by an AI tool.
With all of this in mind, let me quickly recap what we learned and highlight some takeaways. First, engagement with AI is higher. The publishing industry has become more engaged with AI, but not necessarily more enthusiastic about adoption. Second, AI is improving efficiency. Seventy percent of AI users say AI improved their efficiency or productivity. And we noticed that in the majority of use cases shared by their respondents, AI was used primarily as a time-saver tool. And third, we learned that staying informed is more important to respondents than accelerated adoption.
Another takeaway is that there is alignment between the Canadian and U.S. markets. We believe that this creates an opportunity to collaborate on joint initiatives and tackle some of the needs that we identified in the survey. This does not mean that Canadians should ignore calls to act on matters unique to our market and workforce, where data sovereignty and independence have been a widely discussed issue in recent months.
In addition to this, we learned that addressing trust is a priority. Throughout the survey, respondents from different parts of the industry highlighted issues around trust. Some of them related to not trusting the tools or the companies that develop them. Some related to not trusting their colleagues, partners, or collaborators to be transparent about their use of AI. And the list goes on.
And the final takeaway, the top concerns remain. While adoption increases in some areas and new applications arise, the top pain points and concerns remain top of mind for many respondents.
So, that’s all the survey data I had to share with you…I had to share. But before we move on to the Q&A, I wanted to flag some additional resources BookNet has available that are related to AI and you may be interested on.
First, in our Tech Forum Content Library, we have recordings of sessions covering marketing, metadata, examples of applications of AI in the industry, and more. So, I encourage you to browse the library and get access. We share transcripts, we share slides, and we share recordings. So, there’s options. Next is the BookNet blog, where we share and highlight standards developments, including EDItEUR’s aspects of AI in ONIX documentation, which is crucial for anyone wanting to disclose AI used. And we also share news about the fielding of the survey and the publication of the report. So, if you want to stay on top of that, that’s a way to do it, among many other topics, including research and other things.
And that’s the end of my presentation. Thank you so much for your attention. Now, I’m going to welcome Lauren. Hi, Lauren.
Lauren Stewart: Hi. Thank you so much. Hi, everyone. I’m just going to let Nataly get ready for the Q&A, get some much needed water, stretch it out a bit. Just wanted to thank everyone who did reply to the survey, as well as all of our industry partners who helped communicate and bring some awareness to it. We had a great report writer in Jarin Pintana, and also wanted to thank the BISG team for their continued partnership on this survey, as well as many other initiatives that we work on with them.
Also want to encourage you to reach out in case you have any suggestions or feedback on the survey. We are actively thinking about 2027 and what would be needed there to support you in your work. We would look to be…we’re looking at summer of next year of 2027 to field that. And you can reach us at [email protected] for that.
If you have any needs to stay on top of all things BookNet, including getting updates on slides and videos from our Tech Forum series, including this session, you can subscribe to one or all of our newsletter at booknetcanada.ca/newsletters. I’m also just going to drop another link in the chat for the full report. While you’re waiting for the recording, you can take a mosey over there to take a look at that.
And I’m just going to start with some of the questions that we’ve received. You can continue to drop those into the chat if you’d like. We also have a ton of questions that were submitted during the registration process. We’re going to try and hit as many of those as possible.
And the first one will just be a kind of a reiteration of how Nataly ended her session there. A question about metadata or bibliographic data standards to document the presence of AI content in published works and how publishers can help consumers identify AI-generated or AI-assisted works. And for that, I’ll again point to this paper that was done by EDItEUR. It’s called Aspects of AI in ONIX. It’s what they do through their format of an application note. They release these regularly. There’s probably about at least four year, as many as maybe six a year. And this one is getting updated continually as not only does the ONIX standard change, but as more and more applications emerge as we talk about AI in our industry. So, that link is in the chat. It’s a great resource.
And Nataly, if you’re ready, let’s start moving through some of these questions. First question here is how Canadian publishing is adapting to AI. And how do people working in Canadian publishing see its role in the future? What do you think about that one?
Nataly: Yeah. So, that’s a good question. So, based on the data…check the data to answer this question. About half of the organizations the publisher employees work for are using AI. And the areas where they are using AI tools the most are marketing, metadata and title optimization, and administrative. So, very close to what we saw for all Canadians in the presentation.
As for the future, the closest thing I can offer is we asked AI users working for publishers, of…well, everyone that were using AI, but filtering by publishers, we asked about the new areas where they anticipate organizational use. And over half of respondents were unsure. And the next choice was production, layout…so, layout design, typesetting, file preparation, etc. That’s where they anticipate use.
And we also posed the same question for those who are not currently using AI at their organizations. And their response was similar. So, over half said they were not planning on using AI at all. And the next most popular response was marketing.
Lauren: Great. Thank you so much. Another question about Canadian publishers and looking at how they relate to their peers in the U.S. industry. How are Canadian publishers? Where are they in relationship to their U.S. peers in adopting AI?
Nataly: So, there are different ways to look at this. But in terms of reported AI use by publishers, it is slightly higher in the U.S., but the difference is only 16% between the two markets. So, it’s quite low. If you look at this in terms of who is in an exploratory phase, we found that for both markets, 23% of respondents said that they are approaching AI carefully and just exploring what’s out there in trying to figure out a definitive approach.
Lauren: Yeah. That seems fair. And I think that that kind of tentativeness is something that touches on what the next question we’ve lined up here, and something that you kind of explored at the end there when you were talking about trust, which I think is really important when we think about AI. And this question was worded really sensitively, how to handle polarised feelings around AI respectfully. So, they’re probably thinking about that, either as someone working for a company or even as a leader. What’s your thought on that one there?
Nataly: For this one, I’m going to refer to my notes because it’s a big one. But the first thing that I want to do was acknowledge that the way most of these technologies were developed was terrible, and had no consideration for creators or for those who will be impacted by it. So, naturally, there will be a strong and valid feelings that people are going to have, especially in our industry. And now that the technology is being used and available to everyone, and that in many cases, employees feel it’s being pushed from the top down, I believe that the best way to ease those feelings is by nurturing and strengthening trust. And this has to be done across the book supply chain. No one can escape this.
I’m actually in the very, very early stages of developing plans for a webinar about this subject. But so far, what I believe can help with this is, first, making the creation of AI policies as… Oh, my god, this is a hard one for me. Where everyone can contribute.
Lauren: Yeah, collaborate.
Nataly: Yes. And making sure that there is dialogue and consultation within the organizations. At the end of the day, they will be the ones using the tools. Another way is making the protection of IP a top priority, and working with authors to find ways to be respectful of their work. Some may be okay with licensing, some may be not. I think, transparency, it’s just the best policy when it comes to this.
Another thing is working on adopting metadata centres that help communicate AI-related information. So, staying on top of any of the developments that you were mentioning earlier is very important, and not just staying on top, but making sure that, A, that information is getting added to the metadata, and making sure also that the systems can take this information from the data centre to the data receiver, and that the data can be displayed. So, there’s a lot that has to happen here. But if adoption increases, then all these things will start to happen.
As you can see, for me, I think that everything goes back to being transparent. So, being transparent with your colleagues, with the creators you work with, with your vendors and partners, and ultimately with readers. I think that polarising feelings can be easier to handle when there’s evidence of thoughtfulness and a desire to work together to figure out these big challenges.
Lauren: Yeah. I think you’re right there. I think we’re dealing… When we talk about trust, we’re also talking about fear. And I think there’s so much fear happening here. There’s fear from what we’ve seen as copyright challenges from creators. We are seeing fear in terms of job stability for freelancers, as well as people that are in salaried positions or creative positions working in publishing. There’s a lot of concern here that we’re going to have a stable industry with jobs. And I think employees actively feel that.
And I think another area where this kind of touches on, and I’ll ask this next question is, do you think publishers will enforce transparency among their authors or illustrators and promote that information? This is another question from the registration.
Nataly: Yeah. Well, based on the data alone, we can’t confirm or deny whether this will be implemented. But one thing that we learned through the survey is that over 70% of publishers are implementing or working on establishing strategies to safeguard the IP of creators. So, to me, that’s a really good sign.
Lauren: Yeah.
Lauren: Yeah, that they’re working on something to protect their creators. And I think it also implies that there is some thoughtfulness being put into that relationship. And I know we’ve talked a lot, like you and I and our work about kind of power dynamics, and that there’s a big piece of this as you’re working with your contributors, with authors, with illustrators, whether they be in-house or freelance, to ensure, going back to that collaboration, that you’re understanding how they’re using AI, does that align with your in-house or your company’s intended uses of AI, and ensuring that that communication is updated, that it’s free flowing, and that it’s about seeking information and not trying to necessarily judge, but gathering as much information as possible.
Nataly: Yeah. And…
Lauren: Oh, sorry. Go ahead.
Nataly: I was just gonna add something. I think that the market is also demanding this transparency. Libraries want to know whether the books that they are planning on buying were AI-assisted or AI-generated, or there was no involvement whatsoever. Retailers want to know also, and readers want to know as well. So, there are many reasons to work with authors, and together, come up with ways to disclose this information. And there’s also the metadata element that we were mentioning earlier.
Lauren: Yeah. That aspects of AI and ONIX paper is excellent to kind of get started on disclosing that in the supply chain in a systematic and really thoughtful way. I do recommend that for people. Another question, kind of pivoting, and this might take you back a bit to…you mentioned some presentations that were done in September on Tech Forum. And so, we can put those links in the chat again. I can grab them in a second. But how can we increase our visibility for our books on AI tools such as ChatGPT? And this was a question from registration.
Nataly: Yeah. So, this is changing very rapidly. And with ChatGPT, for example, now having the possibility of advertising, we just don’t know how the information will be prioritised anymore. And there’s a lot of unknowns, if you will. But in September, we hosted two wonderful sessions with amazing speakers that addressed this issue of getting discovery in the era of AI through two angles. So, one of them was through metadata and strengthening metadata, and the other was through marketing. I’d recommend watching both sessions. They are really good, and a really good starting point to implement some strategies that will allow you to compete.
Lauren: Yeah. They’re a great little pair together. Fantastic work by those speakers. We’ve put the link in the chat. I would definitely recommend people check those out as you consider how to use ChatGPT and search to discover up your titles. Another one from registration that I think kind of pivots us, and we can look at. You’ve mentioned libraries at several points throughout your presentation. And this is a question here, how best to incorporate protecting the integrity of publishing and collection development in libraries?
Nataly: Mm-hmm. Another big question, but I love it. I feel like I’m just saying the same over and over…
Lauren: It’s a little bit weighted, for sure. Yeah.
Nataly: No, no, no. And I’m happy to take it, but I think it all goes back to transparency. So, authors need to be truthful about their use of AI, publishers need to do the same in addition to disclosing this information. And then the systems need to be able to display their information, as I mentioned earlier. Because library buyers, for example, they want to make informed decisions. So, that’s one way. And I also believe that this has to be approached as an issue that is constantly changing, even from the technology perspective and the policymaking perspective.
So, in addition to what I just mentioned, I believe that it requires continuous learning and staying informed on what goes on in other areas of the book supply chain. So, for example, if you work at a library and you are here learning about what’s happening in all the other areas, that’s great. I think that that really helps with an understanding of what’s happening, and then you can make some decisions for your own context.
I also believe that collaboration is a great way to start developing strategies. There are, for example, groups of library employees working in collection departments who gather virtually to discuss issues, to strategise. And this community approach, this peer approach, I think, can be really helpful when addressing these big questions.
Lauren: Perfect. I think we’ll just…I’m going to just move to another question, because you’ve now…we’ve talked a bit about publishing, a bit about libraries. And in the survey results, you did have some insights from bookstores. Would you mind kind of sharing some of that information there? How are booksellers leveraging AI? Are they?
Nataly: I really wanted to know more about this, but we only got 17 responses, which is still amazing. But for next year, we’re going to make sure that that number increases. So, what I’m going to share, it’s not representative of the whole market, but I think it gives us a really good indication of where people are at. So, the majority of respondents said that they were avoiding or encouraging others to avoid AI. And that’s where they are at, the large majority. I’m talking like 16% of them. Only one of them said they were using AI, and it was for marketing purposes.
Lauren: Okay. Thank you. I’m conscious of time. So, I think we’ve only got time for one or two more questions, if you still have got it in you. This is an interesting one. How are people in the industry that are committed to environmentally conscious choices reconciling their mandates with AI’s ecological impact?
Nataly: So, we didn’t ask that explicitly in the survey. But based on the data, we can highlight that the negative impacts on the environment and/or the organization sustainability goals was the second most popular risk-related concern among Canadian publishers. So, they are thinking about it. They are concerned about it. And beyond this survey, the Green Book Alliance is working on professional development opportunities about this topic specifically. So, it hasn’t been announced yet. But the best way that you can find out about what’s coming up, it’s by subscribing to their newsletter. But it is something that they are considering. And we may maybe add one question or two following up about how are they thinking of sustainability and reconciling those things.
Lauren: Yeah. That’s a good point there, how we are dealing with people’s individual feelings and how they reconcile against these kind of larger issues that are happening not only societally, but also that represent kind of challenges to a company’s identity as well as the employees working there. And so, I think I want to close us off with a little…one I think that I can see a trend line through your presentation. There’s a question from the chat today, or from the Q&A, I’m sorry. Why do we think engagement with AI is higher while enthusiasm about AI has decreased? Do you have any thoughts there about why we’re seeing that tension?
Nataly: Yeah. So, in the survey, at the very end, we had an open field where people could share anything they wanted in terms of how they were feeling about AI in our industry. And something I noticed in their responses is that a lot of people were saying about how it’s not them wanting to adopt these technologies and more so leadership pushing them to, like, okay, we need to start adopting AI, we need to start doing something here in this space. So, that can be a factor, for sure. And then there’s the enthusiasm about AI decrease. I think that we all know about what’s been coming up in the news. And there’s so much… How do I say this in a way that it’s actually helpful?
Lauren: There’s a lot of political tension here. And there’s kind of a pushback that’s overlapping with a number of issues that are happening in the wider culture. I think that that’s totally fair as an observation
Nataly: Not to mention cybersecurity issues. And there’s so many different layers. And it gets so complex that I totally understand why some people may not be as enthusiastic as they were many years ago, when we were all thinking of AI of, oh, it’s just going to write an email for me, or it’s going to save me hours. Now, we know more, and that may be playing a role here.
Lauren: Well, I think that’s kind of where we’re going to have to wrap up today. Did you want to wrap us up, or would you like me to?
Nataly: Sure. Yeah. Thank you. Yeah. So, thank you. Thank you, Lauren. Thank you to our wonderful ASL interpreters and captioner, and to all the attendees. Before we go, we’d love it if you could provide feedback on this session. My colleagues will drop a link to the survey in the chat. Please take a couple of minutes to fill it out. As I mentioned earlier, we will share the recording of this session as well as the slides and the transcript. And I will email you as soon as it’s available. And lastly, I would like to thank the Department of Canadian Heritage for their support through the Canada Book Fund. And of course, thanks to all of you for attending. Thank you.
Lauren: Thanks, Nataly. Thanks, everyone.
Nataly: Bye.