As I dive deeper into the complexity of artificial intelligence, I find myself hassling with a seemingly benign inaccuracy in my thinking which I cannot, for the life of me, shake. The focal point of this bothering thought concerns the difference between artificial intelligence (AI) and machine learning (ML). Learning about AI, from the beginning, I had immaturely grouped ML in all the other jargonistic groups: deep learning, neural networks, transformers, GANs, the list goes on and on. Recently, however, I have noticed a stark shift from people in the industry — maybe it was always there and I was blind to it, perhaps not. Regardless, they seem to keenly differentiate — strictly — when they are talking about artificial intelligence or machine learning; but never in the same breath.
My first thought was, Who cares? When I wrote my first book on AI for Gen Z, I made an effort to group everything under the umbrella term artificial intelligence. To me, this would help an outsider build their on-ramp into the vast, complex technology. In retrospect, I see my blatant mistake. AI and ML are parallel in thinking but the tracks of thought are separated by miles. Artificial intelligence is a journey, machine learning is but a mere step along the way. In fact, even this is under-selling the difference.
Artificial intelligence is a human experiment in creating non-human intelligence from scratch. Machine learning is a root of this experiment that became its own organism entirely. AI involves meaning, ethics, and evolution in a conquest to mimic human intelligence in an inorganic, unnatural, artificial format. ML involves statistics, pattern recognition, and self-improvement.
In the real world, most of what we talk about as AI is really ML. Social media algorithms, Boston Dynamics’ robots, and Google Assistant, to name a few, are simply machine learning tools designed to improve your experience and themselves.
Thus, the final question must be asked: Where is artificial intelligence? Where can I point to and recognize, “Ah, that is artificial intelligence! Not ML, but AI!” For what it’s worth, there is no artificial intelligence (that is widely circulating and available for public curiosity). AI has become a buzzword of sorts, when in reality, machine learning is what 9/10 people are referring to. Right now, AI can be conformed into the uncomfortable but radical term “AGI” or artificial general intelligence.
Another can of worms to be opened at a later date, the basic gist is achieving human-level intelligence, or general intelligence, in a computer. (We can create AI’s that drive cars and play chess — these are forms of narrow intelligence. But the AI in my car cannot play chess and vice-versa, hence the narrowness of the intelligence. [You may also note that I opted to use the term “AI’s” to explain these narrow intelligence’s, when they really are machine learning tools.] Creating a form of computer intelligence that can drive a car and play chess with the same tenacity would be, in the most convoluted sense, artificial general intelligence.)
Phew. Glad I got that off my chest. Hopefully, through this blog post you can and will start seeing how misguided most people who talk about artificial intelligence are — this includes myself. Let me be clear: there is nothing wrong with this shift and fixation on machine learning. However, being clear on what we’re talking about and what the differences are can be of substantial benefit, for we are still trying to emulate human intelligence, putting our best qualities in machines and hoping, praying, that some good will eventually come out of it. Let’s hope it does.