I like to ponder about the next big thing. At the same time I marvel about what has come to pass, such as the personal computer, the internet and smartphones. One common theme I have noticed is that whatever the next big thing is, by the time I realize it, it has likely been here for a while. Technology generally doesn’t just appear but instead takes years to develop, improve and gain popularity.
AI seems poised to be that next big thing, if it isn’t already. Adding some level of intelligence to many of the devices we use and own as well as services we use seems like a natural fit.
The term AI currently covers a range of technology, primarily related to machine learning. This relies on the idea of training a computer utilizing sets of data to identify something. You could teach a computer what a sailboat looks like by showing it many pictures of sailboats. The computer is allowed to create its own algorithm for what matches based on the training data, which it then uses to identify unknown items.
This same methodology is used to teach a computer to act on objects it recognizes. Examples such as adding cartoon eyes to pictures or swapping faces are common. More complicated tasks like knowing when to stop a car when someone steps in front of it, or the nuances of speech and language, are next. All of this is taught by presenting thousands of examples and letting the computer build the algorithm.
What the Future Holds
What all this means for the not-so-distant future is anyone’s guess, but I will throw my hat in the ring and say it is going to be a little of everything for a while. For starters, the devices we use every day will become more intelligent. The intelligence will primarily live in the cloud, so at first this will mostly take the form of access and integration with existing devices like smartphones, automobiles and voice assistant devices. Over time, other technologies related to augmented reality (AR) and various internet of things (IoT) applications will come into play.
At first, this may be your smartphone with an AR overlay feeling more like a friend walking with you while giving you directions and information. It could be calling or texting your fridge in natural speech to ask what you are running out of while you are at the store. The interface will be more natural feeling (and possibly creepy) but the goal and task of getting us where we need to go or making a shopping list will be the same as they have been. At this point, some of us may feel creeped out knowing that we are talking and interacting with a computer like we would with another person, while others will like the new interface. AI will continue to make its way into just about everything we do in many cases without us acknowledging it; in many ways it already has. It will help us write our emails or papers by guessing what we want to say based on what it has learned about our personal and past writing style. It will help architects and engineers design and push boundaries while the AI can pay attention to the less interesting details such as code compliance and structural loads.
It Will Learn From Us
As all of this AI creeps into our everyday lives in ways that I think will mostly be accepted or at least tolerated, behind the scenes it will be learning. Our use will become a larger and larger data set to improve the training and performance over time. At some point, the AI that has been assisting us will be good enough to start working on its own. This won’t be a sudden change to autonomy, just that it will get better little by little.
Eventually, all the jobs that AI has been helping us with could be done by AI, which will start with some human oversight. For example, instead of writing an article, the AI will write it and a person will review and edit it. Over time and with more learning, it will likely be possible to eliminate the human oversight if desired and trusted.
Testing autonomous vehicles with a driver in the seat to take over is a good example. While the AI can do 90 percent or more of the driving, there are still rare cases where it fails and a human driver is needed. When an accident happens (tragic as it is) it will be studied and data-integrated to make sure it can never happen again. Ideally, AI would do the job with zero accidents although realistically it simply needs to be better than people are.
If AI Takes Control
At some point, AI may learn enough to take on the jobs it has assisted with, but this will probably not happen all at once. I don’t think people will want to give up the jobs they are doing, and many will feel that the AI can’t do as good a job or that said job needs some imperfection (a human touch). No matter the case, this will be a strange time that is probably closer than many of us think. Many have already thought about this in books and theory, but it will bring these discussions to the forefront. Questions about the future of jobs, and where we should draw the line. Questions about AI consciousness, paying AI devices and the concept of AI slavery also exist. Likewise, do we need to fear AI as we give it more control over our lives?
I doubt that these changes or questions will happen all at once, but they will happen. This will simply be technology advancing, learning and slowly doing more over years until one day we look back and wonder how we got here.