AI vs Algorithmic learning: What's the Difference?

     Lately everyone, and I mean everyone seems to be claiming to use AI. Are they, though? The short answer; no. Let's expand that.

    

    Above you see a poster for the movie AI: Artificial intelligence. The movie is about AI. The definition of AI is clear: AI is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy.

    That last bit about autonomy is key. Both in the film and in the definition. The last part means that once you build it you no longer have any control over what it does, or doesn't, do. This is why Asimov made his rules.

They are:

(1) Don't HARM HUMANS 

(2) OBEY HUMANS 

(3) Protect robots unless it conflicts with rules 1 or 2.

I am paraphrasing for conciseness. IF we think back to the definition of AI, it can learn and it has autonomy. This means it can decide to ignore it's programming. Basically Asimov's rules exist to try and nullify AI.

    Can today's technology do any of this? It CAN solve problems depending on complexity. IT CAN'T do any of the rest of the list. It can't think for you, it isn't creative, it can't decide things for you, and it doesn't have autonomy.

 

   Algorithmic learning, in the context of artificial intelligence, refers to the process where machines or algorithms learn from data to improve their performance on a specific task.

 

   So algorithmic learning is best illustrated in action by Alan Turing's ENIGMA machine used during World War II. This complex beast of a machine was programmed to try and crack codes used by nazis. The machine required a human to program it. 

 

   It ran day and night trying all possible combinations of letters and numbers to break this code. On it's own it may have taken decades to break the code, had someone not realized the same length grouping of letters was always at the bottom of each message. Once a human figured out those words. they could program ENIGMA to break the code every time it was changed.

    While algorithmic learning is the foundation of AI, it is not AI. Not by a long shot. Could you imagine what may have happened if ENIGMA had autonomy and the capacity for decisions. It may have refused to do it's job, started making up answers to spite its' bosses, or maybe even given bad Intel due to sympathising with nazis.

 

   It's a very good thing all it could do was solve a problem. Today's algorithms have complex capacity to solve what they are programmed to solve. That is not AI. Not yet.