Google recommended a recent publication in NYT about their Artificial intelligence program (A.I.) Going out on limb they used machine learning to generate the recommendation. Find this to a tad ironic, but glad to have “stumbled” across it.

One of the key paradigms of Intelligence is our ability to process natural language. There are 5,000 languages spoken in the world. English contains about 490,000 words and another 300,000 technical terms. The average person’s vocabulary contains about 60,000 – 100,000 words. When you factor in semantics and syntax, programming a translation service becomes immensely complex.

Luckily Google has “Chuck Norris” level engineers – in particular Jeff Dean (his PIN number is the last four digits of pi). With the help of Jeff and other engineers the Translate service has improved dramatically. Most of these gains occurred overnight when they switched from a classically programmed engine to an A.I system. Google Translate is just one of many A.I. applications. e.g. automated call centers, fraud detection, recommendation services, and autonomous vehicles.

I wouldn’t worry about losing your job to A.I. just yet. Humans have two primary advantages:

  1. Our Brains – How it functions is still somewhat mysterious. It’s got billions of neurons each with a freakish amount of connections and computing power.
  2. Our Eyes – Machine optics and image recognition are improving but still weak. You can try out image recognition services from Microsoft here - https://www.captionbot.ai/ If you play around with it enough you’ll understand my “improving but weak” comment. If you do play with it be sure to provide feedback in the form of rating. Each rating will make the program better in theory.

It’s possible that intelligence cannot be completely described. If that’s the case, then A.I. research might never reach the science fiction levels. Hopefully one day A.I. will change the world for the better e.g. cure cancer, free people from mundane tasks, or eliminate the language barrier.