Google Brain AI is working on Wikipedia articles
Internet is huge and widely spread and has tons of data in form of articles, blogs, pictures, videos, tweets and much more. It’s impossible to keep up with such huge data. An AI with a better algorithm in future can certainly make it possible to track, read, categorize and put it to a better use.
The race to perfect the AI has been taking a new turn everyday. Right from Google, Microsoft, Intel, Facebook and many more are working and collaborating in different segments to perfect the system and get the desired results. Alexa of Amazon is one of such AI which is having a moderate success so far.
Google Brain has been working on a process where it’s trying to train an AI to write Wikipedia articles. Although summarizing articles on web pages are not easy and they had a varying degree of success.
In a paper, out last month which is accepted for this year’s International Conference on Learning Representations (ICLR) in April, describes difficulties of text summarisation.
Just like Google, Salesforce has been working on a model to create relevant and highly readable multi-sentence summaries of long text, such as news articles, significantly improving on previous results and their results weren’t bad.
However, the results generated by these computers are short and simple and they lack creativity and rhythm of text written by a human. Google Brain’s efforts are better as their sentences are longer and seem more natural.
Although Google Brain had a fair amount of success, it’s a long way for it to develop effective text summarisation or generation. At this point of time it seems unwise to put it in use for a system like this to auto generate Wikipedia entries.