AI versus Library Databases

Google Scholar, for our more serious researchers, is a “go- to” place for information. It is so much easier to use than searching a database with Boolean operators. Researchers gravitate to the quick-and-easy solution. I tell researchers that Google Scholar will generate thousands of results, but when going to a database, their topic can be whittled down to a mere 20 relevant article. Further, I explain to researchers that they are not only searching resources on their topic, but also thousands of other topics that are not relevant.

How do you determine which article to choose in Google Scholar? Google Scholar can be linked to the researchers’ library holdings, or they simply request articles through their library’s interlibrary loan service. If Google Scholar suggests 100 articles, will researchers want all 100 or scroll through the results to identify the 20 relevant ones? The time it takes to construct a thoughtful Boolean search will, in the end, save time for the researcher, but it’s hard to convince them of this. They’d rather cut back on the time it takes to build a complex search strategy with Boolean operators and filters in order to find those 20 relevant articles. It is difficult for librarians to convince researchers to use databases when they can simply search Google Scholar or ask Siri for the answer. There is no need to go to a database to retrieve a biography when I can ask Siri, “When was William Shakespeare born?”

Of course, Siri and Alexa AI assistants have their issues. As an iPhone owner, I use Siri to make hands-free phone calls when I am driving. Apple’s attempt to make Siri seem more human than like a robot resulted in some snarky replies from Siri when I asked it to call my mama. I once asked Siri to call “mom,” and Siri replied it did not know me or my mother!

Michael Simon writes in Macworld , “Siri really needs a personality boost. Apple has added some fun interjections when asking questions and its responses are very natural, but it doesn’t play games, read stories, or tell jokes nearly as well as Amazon Alexa or Google Home” (macworld.com/article/3257605/ios/siri-vs-siri.html).

Another issue of concern is the source of the information on which any of the personal assistants base their responses. Frequently, it’s Wikipedia.

CONVINCING RESEARCHERS TO USE DATABASES

The ease of use of Google Scholar, added to that of the voice-activated personal assistants, makes it difficult to convince researchers to use the databases that require complex search strategies and training to conduct the most effective search. Companies that provide databases must improve the search functions of their databases and make them easier for researchers to find what they are looking for more quickly. It is expensive for libraries to purchase these databases. The de crease in use from our researchers makes these subscriptions a hard sell. Why should libraries continue to pay millions of dollars for subscription databases when our researchers are using Google Scholar to conduct research?

Libraries could instead save millions of dollars by discontinuing subscriptions to these databases, and focus on teaching researchers how to effectively search Google Scholar or how to use Siri to find biographic information for your researcher paper instead of how to develop an advanced search strategy using Boolean logic to find literature in a database that is complex and difficult to use. But that could be a bit extreme.

Matt Enis points out in Library Journal that library vendors such as OverDrive, Demco, and EBSCO have launched ser vices that will work with AI assistants “enabling library patrons to ask Alexa and soon, other assistants to list upcoming library events, get directions to their local branch, ask about operating hours, place holds, or even check out and read borrowed audiobooks” (libraryjournal.com/?detailStory=voice-activated-technology-focus). Libraries are being presented with new opportunities to provide additional services using AI assistants.

Jumping on the AI Assistant Bandwagon

Many libraries have already made content available via AI assistants. In “Flash Briefing,” mentioned earlier, Michael Stephens also discusses his experience using Alexa. He added Alexa skills from the The Los Angeles Public Library “to find hot fiction and library blog offerings”; Alexa will even read the library blog posts.

Bret Kinsella, in Voicebot, writes that third parties are using voice assistants and making their products compatible. According to Kinsella, “Amazon announced at the end of August that there are 50,000 Alexa skills  worldwide and 20,000 Alexa compatible products. The compatibility figure has grew 5x in eight months. Google followed quietly by updating their figures to 10,000 Google Assistant compatible products, a seven-fold increase in nine months” (voicebot.ai/2018/10/19/phase-one-of-the-voice-assistant-era-is-over-long-live-phase-two). These numbers have probably increased since the August 2018 date that Kinsella cites.

Privacy Concerns

With every new technology comes some concerns. One of the biggest concerns with using AI assistants is privacy. Libraries are often willing to take on new technologies without taking in to account things such as privacy. When I heard on Good Morning America how a Portland, Ore., woman’s conversation was recorded by her Amazon Echo smart speaker and shared with her husband’s employees in Seattle, my heart dropped. Niraj Chokshi writes in The New York Times about this story and discusses how “researchers at the University of California, Berkeley, said in a published paper that they had proved that the technology could be exploited, too. The researchers said that they were able to hide commands in recordings of music or spoken text that went unnoticed by humans but were understood by personal assistants such as Apple’s Siri, Google’s Assistant and Amazon’s Alexa” (nytimes.com/2018/05/25/business/amazon-alexa-conversation-shared-echo.html).

I do not send personal emails at work for fear I will be hacked and my most inner thoughts will be shared with my co-workers or, even worse, my boss. Can you imagine intimate and personal conversations with your loved one being shared with your co-workers? Conversations in my home are private and are only for family consumption. These types of privacy concerns should be considered by libraries as they introduce AI assistants into their library programming and work.

AI assistants have made my daily life so much easier. I can easily multitask. I can talk to Siri or Alexa as I do my laundry. Our researchers are having the same experiences and want finding information in the library to be just as easy. It does not matter if you are a high school student writing a report on George Washington for history class, or an entrepreneur looking up company information to make a business decision. This information is easy to retrieve using AI assistants, and the library can work to make finding information so much easier.