In the recent past, i had written saying that we will move in the near future from "search" engines to "do" engines. Looks like Eric Schimdt recently has been talking about serendipity engines as the future.
Forrester writes about this.
As you look into the future, the distinction between “search” and “discovery” gets muddy. While it sounds like science fiction to suggest that technology can help search for things you don’t even yet know you want, the opportunities to improve human discovery are very real. Combining a person’s context—where they are, who they’re with—with their past opinions and actions and the opinions and actions of others can create tremendous value and relevance.
Imagine you’re someone who has positively rated Mexican restaurants in the past. As you drive through town around lunchtime, your device alerts you to a well-rated Mexican restaurant that is nearby and likely to suit your tastes. This information may not be welcome at 8 a.m. or 2 p.m., nor would it be welcome to someone who hasn’t expressed an affinity for Mexican food. It is the combination of social media, individual preferences and context that creates the opportunity for proactive discovery rather than reactive search.
This isn’t about opening your Yelp application on your smartphone and seeing the same search results as everyone else; it’s about having hardware and software that intuits and presents the things you really care about. There are already examples of simple "Serendipity Engines" available, such as Netflix's rating system—the more movies you rate, the better the recommendations you'll receive.
What is exciting about this future is the continued progress toward empowering consumers.
How will this evolve over the next few years will be interesting to track. Service & solution providers will need to build and integrate their development around these trends.