Now that we have covered basic techniques for recommender systems, choose one commercial recommender and describe how you think it works (content-based, collaborative filtering, etc). Does the technique deliver a good experience or are the recommendations off-target?
As we kicked off the interesting topic of recommender systems this week, I spent some time searching for information and found that a lot is happening in the corporate and university tech departments around understanding of humans by the software or machines using AI.
A start-up from 2010 in the field of AI called DeepMind was started by Demis Hassabis from London, UK. He and his team of computer and Neuroscience scientists have come up with intelligent models and algorithms using reinforcement learning technique as the core of their works. Hassabis in one of his lectures said “The core of what we do focuses around what we call Reinforcement Learning. And that’s how we think about intelligence at DeepMind.” - https://mondaynote.com/deepmind-could-bring-the-best-news-recommendation-engine-fc66051cf2ca
Google (Alphabet Inc.) acquired DeepMind back on 2014 for half a billion dollars. It is not a shiock or surprise as to why Google is investing so heavily in AI, and only Google can do it because they have the big data from users across the globe to understand and make intelligent recommendations and predictions out of it. DeepMind is trying to create a intelligent engine that would have the capacity to think like humans in a given environment. Environment, observations, model, predictions and actions are some of the keywords often used by Demis. And according to the author of the above cited article, these keywords are the foundation for a good recommendation model or engine. For example a News feed that is designed based on a specific users preferences and likes.
“Personalization doesn’t work very well. It currently sums up to averaging the crowd as opposed to adapting to the human individual”. Another quote from Demis.
My Google news feed is based on my location for local news. I see my home based local news alongside the news associated with the place I just landed or arrived at. This is the easy part, however Google aslo pushed relevant or personalized conetent to me based on my search and browsing habits. I use Gmail and google has full access to my communications including the things I like and prefer, for example a ticket I purchased for a concert or a music band. My feed now includes any latest news or updates about the band. The google AI machine is learning a lot about its users over a period of time. So much that their software could most probably find a missing person before the offical agencies can.
“Personalization doesn’t work very well. It currently sums up to averaging the crowd as opposed to adapting to the human individual”. Another quote from Demis.
Google News feed recommendations works for me based on my experience so far. The new investments being made and products that are going to be introduced in the future should make my experience even better. Bots and engines like Google, Watson, Alexa and Einstein are going to be embedded into many devices and hopefully we call can sit on the couch and let the bots take over our lives :)
Here is another interesting article by MIT Technology Review regarding DeepMind and how it is creating or I should say disrupting the way we curently think about AI or machine learning.
https://www.technologyreview.com/s/601139/how-google-plans-to-solve-artificial-intelligence/