Algorithms are all around us and are a major driving force in our day to day activities, everything from the GPS on our phones that use algorithms to predict the most efficient route to get from point A to point B, to businesses like Amazon that use learning algorithms to recommend products you might like, and even the players on your favorite sports teams who were probably selected using a learning algorithm. Virtually everything in today’s world seems to be powered by some kind of learning algorithm.

  Now back before machine learning really got off the ground the only way to tell a computer to perform a specific task was to write out an algorithm that told the computer how to perform it in very specific detail. A machine learning algorithm, also called a learner, are algorithms that create other algorithms. In other words, this allows computers to begin writing their own programs so humans no longer have to. All a programmer has to do is provide the learner with enough data and it will be able to teach itself everything in that data set by writing its own algorithms, opening up bold new possibilities.

  These learners can either teach themselves knowledge or teach themselves certain sets of skills. For example if a learner was provided with a database filled with the rules of the road, and an ability to respond quickly and appropriately to different novel situations it would be able to effectively drive on public roads without causing any accidents.


  The key to a successful learner is data. The more data you have the more there is to learn. There is more data available than ever before which is in part where the  surge in machine learning is coming from. Big businesses like Google have taken a serious interest in machine learning. Google makes money mainly by selling ads on the web and using machine learning to predict how likely a user is to click on a particular ad, the higher the probability the more valuable the ad. Even a small increase in click prediction means billions more dollars in profit each year. This highlights just how important data is to a successful learning algorithm and in turn the successful implementation of said algorithm.


  If a master algorithm existed then it would be the gateway to solving some of the toughest problems humanity faces. Advances like programming human-like general intelligence into robots or curing cancer will become reality once the master algorithm is discovered. This would be the single greatest discovery in human history.

  A true master algorithm would allow a computer to teach itself how to do virtually anything from any database of information provided to it with no limits as only one learning algorithm  would be required. It would be able to discover the corresponding knowledge from any set of data given to it which means the capabilities of such an algorithm would be virtually  limitless. The idea that an algorithm like that might exist is extremely fascinating to me. I will discuss the evidence for it’s possible existence in future posts.


  How close are we to actually discovering such an algorithm? Well, currently most of the experts in the field believe that it will take at least 30 more  years with few thinking it may come even sooner than that. I believe it will probably take at least 30 years and the master algorithm will more than likely be discovered by an ambitious college student or someone who is just starting out in the field of AI. A younger person will have an easier time examining the problem of creating the master algorithm, because a lot of times older researchers, who have been studying in a particular field for a while have a harder time seeing the forest for the trees.

  Regardless, the door is wide open for just about anyone interested enough in the field to figure out the master algorithm and when that finally happens it will change our world forever.

Leave a Reply

Your email address will not be published. Required fields are marked *