Google's latest AI experiments let you talk to books and test word association skills
Google today announced a pair of new artificial intelligence experiments from its research division that let web users dabble in semantics and natural language processing. For Google, a company that's primary product is a search engine that traffics mostly in text, these advances in AI are integral to its business and to its goals of making software that can understand and parse elements of human language.
The website will now house any interactive AI language tools, and Google is calling the collection Semantic Experiences. The primary sub-field of AI it's showcasing is known as word vectors, a type of natural language understanding that maps "semantically similar phrases to nearby points based on equivalence, similarity or relatedness of ideas and language." It's a way to "enable algorithms to learn about the relationships between words, based on examples of actual language usage," says Ray Kurzweil, notable futurist and director of engineering at Google Research, and product manager Rachel Bernstein in a blog post. Google has published its work on the topic in a paper here, and it's also made a pre-trained module available on its TensorFlow platform for other researchers to experiment with.
The first of the two publicly available experiments released today is called Talk to Books, and it quite literally lets you converse with a machine learning-trained algorithm that surfaces answers to questions with relevant passages from human-written text. As described by Kurzweil and Bernstein, Talk to Books lets you "make a statement or ask a question, and the tool finds sentences in books that respond, with no dependence on keyword matching." The duo add that, "In a sense you are talking to the books, getting responses which can help you determine if you're interested in reading them or not."
The second experiment is Semantris, a game that tests word association skills (while collecting data from users).
The very first thing I thought of to ask "Talk to Books" is "What is a cactus?" The first result is "peyote. A species of small cactus, or the powerful drug decocted therefrom by the Indians of Mexico and the western United States and widely used for medicinal, ceremonial, and religious purposes. G.P.M." Thanks, Google.
Also at TechCrunch.
(Score: 2) by maxwell demon on Sunday April 15 2018, @06:34PM (2 children)
Just from the description, it's easy to find a question that it can't answer: Just ask something which in this combination has probably nobody ever written.
Such as: Did Rincewind visit Hogwarts? [google.com]
Not surprisingly, no useful result comes from this. Although the first highlighted item is surprisingly irrelevant:
(I just note that following the link above gives another set of results, since the question mark got removed from the question. Adding that back manually currently reproduces my quoted result. The problem seems to be SN messing with the link.; here's the original one: https://books.google.com/talktobooks/query?q=Did%20Rincewind%20visit%20Hogwarts%3F )
Any human who has read the relevant books would be able to answer this question: No, he didn't, as both are from different fictional worlds.
As a side note, it's interesting that the Firefox dictionary knows Hogwarts, but not Rincewind. ;-)
The Tao of math: The numbers you can count are not the real numbers.
(Score: 0) by Anonymous Coward on Sunday April 15 2018, @07:04PM (1 child)
It's no fun. I tried questions such as "Is gandalf gay?", "is soylent tasty?" and "who likes dick?" none of which were particularly amusing.
Some results are strange [google.com]
The research may be valuable but I don't see any practical use case for this product in its current form.
(Score: 1) by Ethanol-fueled on Sunday April 15 2018, @10:12PM
It looks like a way to sell books.