Google's Talk to Books: AI Opens Creative New Search Pathways
Google has long been acknowledged as the king among online search engines (does anyone still Ask Jeeves?). As www.google.com became the default “home” page on many of our browsers, it replaced search platforms like JSTOR and LexisNexis, which are now relegated to libraries and educational institutions. It also changed the way we ask questions, consume information, and navigate our own internal mental landscapes.The power that this gives Google (and Alphabet, its parent company) is immense. There is ongoing debate over whether such a monopoly on access to information is healthy, sustainable, or conducive to democracy. As scholars and pundits debate, the rest of us continue turning to Google for answers to queries ranging from “how to fix a water heater” to “what is the meaning of life?”. Now, a new initiative from the company’s experts in artificial intelligence (AI) is once again shifting the framework of access to information online.
AI Re-imagines the Search Engine Experience
Google recently introduced Talk to Books, an AI-enabled search engine demo that answers user’s questions with curated selections culled from the over 100,000 books in the Google Books database. It’s a different kind of search experience for many reasons. For one, it limits results to published works (potentially avoiding much of the false information housed on the internet). Additionally, the AI behind the search field is more advanced than the algorithms that underlie other engines, leading to a more conversational experience.The artificial intelligence component, more than the novelty of searching books perhaps, drives the Talk to Books experience. “How does a computer understand you when you talk to it using everyday language?” Google asks in Semantic Experiences: Experiments in Understanding Language, “Our approach was to use billions of lines of dialogue to teach an AI how real human conversations flow.” In this search paradigm, “the AI is simply considering what (users) type to be an opening statement and looking across a pool of many possible responses to find the ones that would most likely follow.” The results are often enlightening, occasionally frustrating, and always interesting.
Talking to Books
A standard Google search for “what is the meaning of life?” directs the asker to the Wikipedia page for “Meaning of life,” then goes on to provide links to an endless number of other articles, scholarly or otherwise. A Talk to Books search, on the other hand, provides curated quotations from published sources as wide-ranging as What's It All About?: Philosophy and the Meaning of Life, Meaning in Positive and Existential Psychology, and Middle Range Theory for Nursing, Second Edition.This example is somewhat obtuse, yet replaying the experiment with a more straightforward question yields similar results. “How to fix a water heater” yields a wealth of DIY articles when entered into Google, but in Talk to Books provides both helpful (“a quick test you can use to see if your gas water heater is venting properly”) and strange (“investigators must disconnect the power source to the electric water heater before investigation”) quotations from Home Maintenance for Dummies and Child Abuse and Neglect E-book, respectively.In both cases, a user seeking a straightforward answer might prefer the standard Google search model. However, someone looking for a jumping-off point for a larger, more complex conversation is likely to be invigorated by their interaction with Talk to Books. The latter offers a less formulaic, more dialectic experience. As the AI attempts to intuit the direction of the searcher’s queries, it continually refines its responses and the material it presents.
Is Machine Learning the Future of Online Searches?
This process, in which artificial intelligence learns through repeated interactions, is known as machine learning. It’s already had great success in image and video search applications, as AI learns to spot and tag patterns in images that can help refine search results and categorize enormous amounts of visual data for easy access.The ability of platforms like Talk to Books to learn from us as we learn from them creates a two-sided search paradigm. What does this mean for those of us at the keyboard? It prompts us to unlearn some of our Google-ingrained search habits and begin asking more nuanced, intentionally-phrased questions. It asks us to become (for now, at least) more curious in our intellectual approach, to consider wide-ranging sources as possibilities for inspiration, to embrace a dialogue instead of an information binary.Will the market as a whole be interested in adapting to a search engine that doesn’t always give us exactly what we want? That seems unlikely. Yet, for creative thinkers, innovators, writers, and artists, Talk to Books is a refreshing experiment in conversational searching. Eventually, we may even forget that we’re talking to a computer.