In an article published yesterday in The New York Times, John Markoff writes about newest achievements in the development of artificial intelligence software carried out by two independent teams of researchers at Google and at Stanford University. Apparently, this new software is capable of recognizing and accurately describing the content of different, often very complex, photographs and videos.

https://www.youtube.com/watch?v=CiWVlduWDLE

Besides its applicability in searching and cataloging photo and video content on the internet, and developing assistive technology for visually impaired, such a precise visual detection tool can be of major importance for improvements of autonomous robots. With corresponding advanced processing and interpretations of imagery, robots will be able to interpret their surroundings more accurately, thus helping them better navigate different complex environment.

The level of interpretation this software can achieve is almost human-like. Illustrations of the software readings of previously unseen test photos can be found at http://cs.stanford.edu/people/karpathy/deepimagesent/.

Google has previously developed similar software which was able to train itself in image recognition. This time they took a step forward and after a relatively short period of training tried their software on a new set of images. They were surprised to see how fast these machines ‘learn’. Many new car models have special software that can recognize pedestrians or cyclists from data sent by the vehicle’s integrated cameras. All that still doesn’t mean that these programs can interpret imagery in a way that will help them ‘understand’ the situation. But, if we relate these advancements to Eugene Gootsman’s passing the Turing test and Watson’s winning Jeopardy, followed by his later employment as a real-life Dr. House, we can attest that artificial intelligence (AI) is in its full swing. These programs are able to recognize patterns and process them in a way similar to our brain-works. The platform is known as neural networks. Since our brain also learns and interprets the surrounding world through experiences and patterns, we can presume these infants of artificial intelligence will grow up to become fully developed adults leading the new generation towards the singularity.