Or better yet, Deep Learning (“DL”). Recently, Apple stated that they propose to add an Artificial Intelligence (“AI”) chip to each new iPhone. So, what is the difference between AI and DL? Unlike DL, AI requires a human interface to pose a question (vectoring the features). A good example of this would be IBM’s Watson, a nearly archaic AI system.
In an AI construct, the computer and software have the ability of intelligent behavior. Once prompted by a human interface or presented with a problem, it can compute one or several responses depending on the question posed. As with all thinkers, the posing of the question is both key and limiting.
In a DL construct, again the computer and software have the ability of intelligent behavior. But without human interface – instead, the computer consumes massive amounts of data, deciphers the patterns (if any), notes the anomalies, and makes observations or proposes solutions to a self-recognized problem. This eliminates the vectoring of features (asking the correct question). A DL construct develops multiple levels of representation and abstraction (organizes the data) and makes sense of the data without all of the layers in intervention and construction. Since DL does not need all the layers of intervention and construction, it is a much faster and more accurate system.
In the near future, we will begin to hear about a company named, Deep Instinct. Deep Instinct is a cybersecurity software, which employs Deep Learning from an Israeli DL firm named 5th Dimension. 5th Dimension grew out of a partnership between Israeli government security agencies and Israeli academia. Two weeks ago, we saw the spectra of the Wannacry ransomware. Ransomware is a virus that requires the computer owner to pay the hackers to regain control of their data. Wannacry hit targets in over 180 countries very hard, and even crippled the British Health Care System. Deep Instinct, where it was employed, recognized the Wannacry attack in micro-seconds, and instantaneously defeated it and cleaned it from the systems being attacked!
This is perhaps the most poignant example of an ideal application of DL. There was no time for a human to ask an AI construct to find out what was happening, then fix it. There are other applications in healthcare, autonomous vehicles, defense, investment, and anywhere else there are huge amounts of data and multiple possible positive or reactive actions that might be taken.
Will DL replace the human being? This is the fear of many and the subject of numerous SciFi movies and books. Among the most intelligent, up-to-date and highly tech savvy IT professionals we have met, the answer is still no … or at least not for a long time. Machines and software cannot take “a leap of faith” and are not creative, because they cannot conceive something for which there is no data. DL currently cannot give us the next Hoola Hoop.