Automation vs. Artificial Intelligence: Know the Difference and Don't be Fooled
As artificial intelligence makes its way into various fields of computing, we find ourselves subjected to a wide-open field of possibilities. And while pimply teens salivate over the idea of Terminator robots reproducing themselves and demanding changes to the Constitution, companies, if they're not inventing them, continue to search for ways to profit from it.
But because artificial intelligence remains an incomplete and unrefined approach to computer-based problem solving, it's not only easily misinterpreted, it's also easily abused. So much so, that since the latter part of 2015, I've encountered more than a few instances in which something was incorrectly labeled "AI" seemingly for the sake of media attention or sales. Here, I want to help clarify the difference so that you aren't duped into paying for a function or feature that doesn't exist.
Automation - It's Definition and Requirements
Automation uses mechanical or technological means to perform a task without human intervention. Wikipedia states that the earliest example of automation is the thermostat, invented in 1620, by Dutch scientist Cornelius Drebbel. Whether or not you agree with its origins, automation has a definite meaning and set of operations.
Today, the robot in the automobile assembly line or food processing plant, just like Drebbel's thermostat and just like First Draft, uses mechanical or technological means to perform a task without human intervention. But automation is not artificial intelligence.
Artificial Intelligence - It's Definition and Requirements
Artificial intelligence attempts to solve problems with computer programs. According to Wikipedia, the idea of performing useful reasoning with a machine may have begun with Ramon Llull (c. 1300 CE). Later, Gottfried Leibniz extended this concept with his Calculus ratiocinator and the intention to perform operations on concepts rather than numbers. Unlike with automation, however, it doesn't yet have a definite meaning and set of operations. It's still evolving largely due to access to better technology. But that doesn't mean we don't have any expectations.
Today, we expect AI to: (1) use native language to communicate with users, (2) store relevant information, (3) use reasoning to answer questions and draw new conclusions from that data, and (4) adapt to new circumstances and/or new information (credit to http://aima.cs.berkeley.edu/). A wide number of video games, mobile apps, and even First Draft do those things and more.
Where It Gets Confusing
Some people believe that the automated selection of random things indicates artificial intelligence. But after taking another look at what automation and AI really are, it's easy to see how that task unto itself is something very, very different. In the act of automatically selecting random things, the opportunity for AI to even exist is gone because the processes eliminate AI's qualifications.
There's no communication exchange, there's no information storage, there's no reasoning, and there's no adaptation. There isn't even a need for it, in fact. The automated selection of random things creates exactly what it sounds like it creates: a set of haphazard, indiscriminate choices. And while there's nothing intelligent about that at all, it's not uncommon to see products that automatically select random things called "artificially intelligent".
Where to Learn More
I don't know why this occurs other than the fact that the interest in artificial intelligence is increasing at a rate that allows for misinterpretations and their exploitation. But as a consumer, it's important to know what they are so that you don't buy something under the impression that it'll do something that it couldn't possibly accomplish -- not even on a conceptual level.
For more about what constitutes artificial intelligence, I strongly recommend following the information in the National AI Research and Development Strategic Plan as well as in other publications by the Networking and Information Technology Research and Development Subcommittee. I recommend this plan because it (a) outlines in surprising detail what AI will mean in the United States and (b) will undoubtedly shape the course of AI's development and its legal ramifications in this market.
Thanks for reading!