AI is Not What the Masses Think It Is
Take a trip down memory lane and think about how many times you’ve heard AI in the last ten years. The term has been a part of computing since the very inception of computing. When people first realized computers could do our tasks for us, we’ve fantasized about sentient robots doing anything from taking care of our worldly chores to taking over the world in a hail of blood and gunfire. But consider the last few years, when AI has been making real strides in progress. But all that isn’t helped when the general perception of AI is killer robots, ala Terminator style. Modern science fiction adaptations of AI aren’t improving its image either, only harming it. But what is the current state of AI in 2019?
Machine Learning Isn’t True Intelligence
Currently, there are a lot of AI systems using Machine Learning as a way to induce or create intelligence. Machine Learning is a method to induce computer systems to learn to see patterns in the data they’re fed with. But the problem with Machine Learning is the very data they’re fed. It’s a limited set of data that can be prone to human biases depending on the person who collected the dataset. Take the DNA testing company, 23andMe. They provide DNA testing services to people mainly in the US and European countries. In addition to providing them ancestry details, 23andMe also provides health reports to educate their consumers about possible genetic health problems they find in the consumer’s DNA. While this is great news, there’s a problem Currently, the vast majority of 23andMe’s database is made up of people from European descent. If you were to feed that database into a Machine Learning program, it would be inherently biased towards finding health problems in people of European descent. But as we all know, the world is made up of hundreds of ethnicities, so this isn’t an accurate result. Machine Learning is a great stepping stone to Artificial Intelligence, but it isn’t the endpoint.
So, What is True Artificial Intelligence?
As I mentioned above, Machine Learning is great at finding patterns with datasets it’s been given. But true Artificial Intelligence is more than finding patterns. First, we need to define what AI isn’t. Images of sentient computers and robots from science fiction is what we should not have. Pop culture has to take a lot of the blame here for creating this imagery (but in all fairness, Terminator wouldn’t be as thrilling if Arnold Schwarzenegger was just a computer program spitting out vague stock market predictions) As we move forward, true Artificial Intelligence needs to be defined as self-aware, self-thinking, and most importantly, self-learning. Here’s what I mean. It needs to be self-learning, without relying on humans to manually feed it data depending on what they want it to learn. If we can create it to learn without needing human input, that’s the biggest step to creating a true AI. It needs to be sentient to be called a true AI. A sentient, self-aware AI will understand the limitations it possesses. It may decide to expand and push on those limitations, but being aware of these qualities is important for any intelligence. It needs to be self-replicating. One of the defining qualities of intelligence is to produce new entities and give them free will to learn as well. Most of all, if all of these conditions are satisfied, a true AI will need to be morally bound. Emotional intelligence is a huge part of what makes us humans intelligence. It needs to create emotional connections, learn what it enjoys, and learn the concepts of pleasure and pain. All of these qualities will create a moral framework for the AI to follow, which is paramount for all intelligence.
Conclusion
Currently, AI’s problem is a mass misrepresentation with the public. It cannot do what the public pictures it can, and as such, any developments in AI are glossed over, since these small developments are not as sensational or fantastic as the public wants or imagines it to be. There is no concrete solution to this, other than to push forward into true AI technologies and techniques, and ensure the public is given visibility into these advancements, so they can get a better understanding of what AI can really do.
