You may not know it, but day by day, we are making huge strides in the field of Artificial Reality. Sure, we are a long way from achieving sentience, but still, we are advancing so fast that current implementations of Artificial Intelligence actually seem feasible.
One way to go about creating Artificially Intelligent Software is to model after the one sentient computer we have: Our brain. By following the way our brain learns, the computer learns too. That’s just theory of course, but we are seeing implementations of this model that are filled with potential.
Wolfram Alpha is one such implementation, as it has gained image processing capabilities, enabling it to theoretically identify any image you throw at it. Now before we move forward, we have to understand how this image processing was achieved. Well, the crux of the concept is this: The software is fed millions of diverse images that cover as much as possible, and the software learns from these images to enable it to recognize other images.
This is primarily how infants too understand their world, as they too process lots of unknown images as they explore their world.
For those not knowing about Wolfram Alpha, it is basically an intelligent search engine that can draw upon data from various sources and can fulfill almost any query you have.
It has the capability to understand Natural Language, eliminating the need to type in a way the computer understands.. This means that it would recognize when you ask it : “Who is the President of the United States?”. Unlike Google, which is basically just a list of links and like Google’s Information Graph, Wolfram Alpha will be able to process data based on that question and will give you your answer.
Then, you can continue to ask: “When was he born” and Wolfram is intelligent enough to recognize that ‘he’ is referring to the person in the previous query.
This is just the tip of the iceberg, though, as Wolfram Alpha can do many things, such as doing complex calculations, finding out what flights are overhead and much more.
Coming back to the image processing features, it does work remarkably well. In most cases, the software’s guess was close to the real thing, and even when it was wrong, it wasn’t too far from the real thing.
We reckon that Wolfram will continue tweaking its algorithms to improve its accuracy.
What do you think? Does Wolfram Alpha look interesting to you?