Google's new search tools aims to search based on meaning over keywords

• Talk to Books is based on the concept of "semantic search", which emphasizes meaning over keywords.
• It answers queries with excerpts from books, which it thinks will be relevant.
• It's not meant to replace regular Google Search.

When you search on Google, the company's algorithms read the words you write and give you the closest possible results. Given the search engine's extensive understanding of what you do online, these results are usually scarily accurate. Yet, Google's search isn't based on the contextual understanding of what you are writing.

AI algorithms are often bad at understanding meaning. Which is what Google's new research platform is all about. It's called "Talk to Books" and it searches through a plethora of books inside Google Books to answer questions.

"Although it has a search box, its objectives and underlying technology are fundamentally different than those of a more traditional search experience. It's simply a demonstration of research that enables an AI to find statements that look like probable responses to your input rather than a finely polished tool that would take into account the wide range of standard quality signals," the company writes on its blog.

Talk to Books was developed by Ray Kurzweil, Director of Engineering at Google. The platform is based on "semantic search", which is aimed at searching based on meaning rather than keywords. It takes your question, understands it and tries to return passages from books that would make the most sense.

It's not meant to replace regular Google Search of course but may come in handy for research purposes. For one, you could easily find books that are relevant to questions that your research is pertaining to.

To be clear, you don't actually get to read the whole book. The tool just picks pertinent paragraphs from various books and puts them in bold font for you to see. And though Google itself admits that it make take you a while to get a hang of Talks to Books, it seems like a good opening attempt. Here's an example.


You can also play "Semantris", a game Google used to train word association to its AI.