Artificial Intelligence: Embrace Evolution!

18/02/2015 16:59 GMT | Updated 19/04/2015 10:59 BST

"The amount of data on the planet will reach 44 zettabytes by 2020"; according to Data IQ. The irony is that that sentence exactly identifies the problem of the field of Big Data - it lacks context. "44 zettabytes" sounds big but doesn't mean anything and for most people it is just unstructured data. Let's look at another sentence (and fact): 90% of the data in existence was produced in the last few years. Here we added 'relevance' by putting it in perspective and that clearly creates the wow factor in our heads. By adding context we created information, which in turn can turn into knowledge when we use that information and make a judgement call next time we make a decision.

Not surprisingly the big technology companies acquired this knowledge and came to the conclusion that Big Data and AI are, well, Big! - resulting in some pretty heavy investments in research. Google bought Deep Mind for $400 million, LinkedIn acquired Bright for $120 million, Facebook more recently acquired the speech-recognition software start-up and the list continues...

But what are they seeing which others might not? Let's create a bit more structure in the difficult to comprehend space of AI and Big Data by explaining the two fields in simple terms. In big data we set the rules for machines to follow. For instance we tell the computer to analyse the shopping behaviour of a particular consumer using a particular algorithm. In AI, also known as machine learning, we let the machine determine the rules based on its own learning - just like you and I would learn.

So essentially the machine decides what algorithm to use based on it's own experience. Take that a step further and we can let the machine question the question. Can we determine any viable information from the shopping behaviour data in the first place? These fields are also known as single, double and triple loop learning. Are we doing things right? Are we doing the right things? and How do we decide what is right?

By now your head is probably spinning and creating fear that if a machine can do that we are all in trouble and that fear is (for now) partially justified. A 2013 Oxford University report indicated that AI combined with robotics could automate half of the US workers ranging from loan officers to taxi drivers to real estate agents and more. Stephen Hawking extrapolated it even further and expressed his fear that AI could end humanity.

Rest assured for now we are safe. The big hurdle rests in what is known as the Moravec's Paradox. Moravec came to the just conclusion that things that are easy to do for humans, like object recognition and perception are extremely hard for computers to do - while simple tasks for computers, such as solving complex theorems are extremely difficult if not impossible for humans to do.

Let me explain with an example. In 2012 Google did an experiment creating a cat recognition system. The neural system was created to recognize 20,000 object categories (not just cats) and that whilst it never instructed what the object actually was, the machine learned the concept of a cat by itself. After feeding 10 million images it nevertheless just managed to get an accuracy rate of 15.8%. This was a 70% improvement versus the previous record but clearly nowhere near human level.

So should we fear the speed of technological progress? We decided to not fear but embrace it. As fear is an emotion based on the disability to predict the outcome of a particular situation, we choose to become better at predicting and finding ways to assist us in making the right decision. Artificial Intelligence should therefore make us smarter.

At Force Over Mass Capital we are in the business of finding, funding and fostering the next big thing. What makes an entrepreneur and a team successful? What is their make-up? Do they have the right constitution of team members? What is their probability of success? We invest in people with an idea, we invest in knowledge and the rest is secondary. Fifteen years after writing my thesis, I have continued a lifelong quest by trying to find an answer to the question, What is the value of knowledge? And now we have gained the right tools for the job, Artificial Intelligence.