For the last 200-300 years there´s something called regression statistics, a regression algorithm which relates known, pre-defined things (today is friday) to know about other things (you are using LinkedIn).
But with Machine Learning we are getting into Bayesian Algorithms, where you don´t need a human to pre-define whats important.
Instead, the computer looks at tons of variables and find hidden correlations. The result is that you can really find all the little details that somehow add up and contribute for someone to open and use LinkedIn today.
In my opinion, the most interesting intellectual challente right now is not in the algorithms themselves, but in finding situations where algorithms can really make a difference optimizing and adding value to that activity or industry.