A version of this blog post first appeared on the Digital Book World site. Bradley Metrock produced the iBooks Author Conference from 2015 through 2017, before Score Publishing acquired Digital Book World, and has authored many articles on the state of the publishing industry and recent trends.
Last Friday, Google unveiled “Talk To Books,” a new tool which uses semantic search – search based on meaning, rather than mere keywords, and powered by the same Google conversational AI used to implement ‘smart’ email replies – to provide an entirely new way to explore books.
What happens when, for the first time in human history, books can be searched at the sentence level, rather than at the author or topic level?
Using this tool, a user can input a statement or a question, and Talk To Books searches over 100,000 books for sentences which best provide a response. No dependence on keyword matching is involved.
This is no quirky novelty Google released that will largely go unnoticed – this is a profound, almost shocking development that will have wide-ranging implications, starting from day one:
1) Book discoverability just became conversational
How do we discover a book? A friend tells us about it. A news source tells us about it. A family member tells us about it. A podcast tells us about it. A website, or website ad, tells us about it.
Now we can add a new one: a conversational AI tells us about it.
As you can tell after spending all of thirty seconds with Talk To Books, discoverability is the problem this tool really helps to address.
I asked the tool all sorts of questions, from “what are some good books about baseball?” to “what are some books about podcasting?” to “what are some books about relationships?” to “what are some good books to read about public speaking?” and more. Each time, the quality of the search results were largely on target, effortlessly producing books I had never read but which clearly spoke, at least in part, to my question or stated interest.
I queried “What are some good books about Donald Trump?” as well as “What are some good books about Hillary Clinton?” and got interesting and valid results both times, which if I were truly interested in learning about either one of those people, would be instant purchases either digitally or in print.
Google’s developer notes explain there is a “popularity measure” implemented which provides a boost to books produced by “professional” publishing houses, but the tool itself is intentionally left unfiltered so as to demonstrate its raw power. My searches consistently returned a selection of books I had never heard of before, all having highly relevant results.
2) Voice-first technology is the delivery system
When people ask Google what books they should read next, they will absolutely not be doing so with a mobile device with a touchscreen, and they certainly won’t be using a QWERTY keyboard on a desktop.
Rather, they’ll be using their own voice, and will speak to a computer.
Thanks to Amazon’s Alexa and Google’s Assistant, and the exponential growth in market penetration of smart speakers and other “voice-first” devices, an entire new generation is growing up with the expectation that you speak to computers first, and if and only if that is not sufficient, do you then get out a keyboard/mouse or perhaps try to touch the screen in front of you. As discussed every week on This Week In Voice, the use cases for voice-first technology are expanding so fast, it’s hard to keep up.
Expect to see Talk To Books rapidly integrated into Google Assistant and made available via Google Home devices, enabling this web tool’s semantic search capability to inform conversational search results delivered via voice. And you know Amazon and Alexa won’t be far behind.
If you believe Amazon has too much marketing power in publishing, you should be a fan of this development, as Google is uniquely positioned to challenge Amazon for bookselling market share. Amazon and Google both have vast data, but the type of data is different between the two companies, and both could be distinctly useful to end users needing to source information or discover new books.
3) What becomes of brick-and-mortar booksellers?
If I can ask a computer, or more precisely, a conversational AI embedded into an object I’m probably already wearing or carrying, questions related to new books I will end up purchasing, and get answers informed by not just my deep personal context held by that computer, but also selected from a massive universe of millions of books, why would I ever go to a bookseller and ask the same question?
I wouldn’t. Bookstores would have to adapt, yet again, and provide something else, because I’m not buying books from them…I’m buying them from the AI that knows my context, that knows the publishing universe, and that can speak to me fluently being armed with both.
4) This development instantly affects all other forms of media
Many individuals and organizations thrive on selling books, while making other types of content, such as podcasts or videos, free in order to cultivate audiences that can then receive marketing to buy said books.
If an AI is telling me what books I need to buy next, that same AI needs to know where to find other content from either that author, that publisher, or similar authors and publishers. And if that content isn’t available in that ecosystem, that might influence the results.
So, in other words, you have to now make sure your podcasts are available through Google Play Music, your audiobooks are available in Google’s ecosystem, and your videos are available and well-searchable in YouTube, if you aspire to sell many books.
When I tell Google Assistant “Google, I know you recommended that book to me just now, but I’m not sure it’s for me” and Google Assistant says “you know, you’re right: in order to make a decision, why don’t you listen to this podcast by the author?” or “here, check out this video, and if you like it, then we’ll buy the audiobook.” then you start to realize the gravity that a development like this has on every other type of published content.
5) Google hit a home run by making Talk To Books’ underlying code open-source
Countless applications can be built using this technology in different contexts, from larger publishers applying this to their own inventories of books to glean various insights, to members of the media using this technology to search for subject matter experts semantically, to schools that could use this technology to expand or update curricula, and on, and on, and on.
Google could have chosen to keep this closed and unavailable for access to third-party development. But they didn’t. It will be exciting to see what gets built on top of this.
And like the core technology itself, we’ll be watching to see how it develops and gets iterated upon over time…and we look forward to discussing all of the developments to come at Digital Book World 2018.