After the tremendous success of Deep Learning, a lot of people were concerned that the next step is that machines take over and humans become the endangered species. Well, the probability that this will happen is not zero, but when recommender’s (“AI”) start to suggest that we should watch Dorian Gray because we also liked the animation movie Wings, the event seems to be very distant in the future. In a previous post, we talked about a possible Recommender Winter which we were not really serious about, but if we count the times that a recommender system actually returned useful results for us, except for relevant products in larger on-line stores, it might have already been arrived.
Since we are also working on personalization and recommendation for quite some time, we know that the work is really challenging and that there are lots of obstacles. However, for companies with more manpower and money, it should be at least possible to come up with something that actually assists users to find what they want. With more and more sites using off-the-shelf components, the situation has not really improved. Maybe because such components do not allow fine-tuning, or there are not enough ratings, or domain knowledge cannot properly encoded, but at the end the results are often hardly useful for users.
In a nutshell, it is time for a Recommender Renaissance.