THE BORING TRUTH ABOUT AI
You can’t turn anywhere today without hearing about artificial intelligence or AI. Most AI coverage seems to exist on a kind of truncated spectrum with two dominant endpoints.
On the one end is absolute AI triumphalism and cheerleading — as in “AI will solve X” with X being any and every social ill and problem that humanity faces and will ever face. In other words, something like the snake oil of the 21st century.
At the other end is the doomsaying of how artificial general intelligence or so-called superintelligent AI will wipe out humanity and end human civilization. Keep in mind that no such thing as general AI even exists, and many experts doubt whether such a thing is even possible. Cautious realists think that even if such a thing came into being, we’re probably a long way off from seeing it realized in any meaningful way.
In between these two extremes is to be found the sensible (or boring) middle. This is the simple truth that AI has been used in many different applications for many years, perhaps in places that people don’t even know about or realize. For instance, personal assistants like Siri, Alexa and Google Assistant have been powered by AI for years.
A broad, workable definition of AI is that it is software that is designed to do a particular task extremely well. Here is the distinction between this type of AI (often called narrow AI) and other types such as general AI. In this sense, AI is not so much a major disruption as another step on the evolutionary path to using software and algorithms to better optimize all kinds of engineered systems.
Manufacturing and automation applications have been using AI in a variety of ways. In particular, AI subsets such as machine learning and neural network tools are used to optimize manufacturing processes by improving data analysis and decision making. They also help to improve efficiency and accuracy.
AI is also finding a home in some motion control applications, particularly those involving robots. For example, AI algorithms are being applied to robotic systems to improve the controller action. Perhaps not surprisingly, AI in motion control and robotics is also focused on the data part of the equation to improve overall outcomes. Gathering and analyzing huge amounts of data can give insights into processes that lead to greater efficiencies, better machine performance, and improved product quality.
Such examples of AI being used in automation and manufacturing appear to be an overall good use of this emerging technology. Of course, not all applications of AI are benign. There are plenty of examples of AI systems making decisions that lead to discriminatory outcomes based on racial or gender biases within the algorithms, which in turn come largely from the datasets used to train the algorithms themselves. There are also issues with the use of AI in facial recognition and surveillance, raising legitimate questions about human rights abuses and the erosion of privacy.
As time goes on, we need to do what we can to promote positive AI applications while preventing detrimental uses.
Filed Under: DIGITAL ISSUES