The good news is that predictions about artificial intelligence have typically always assumed it will be able to do more that it can and that it would affect us sooner rather than later.
Work in the field began in earnest in the 1950s thanks to the efforts of pioneers like Alan Turing, who famously devised the test that bears his name that describes a computer able to mimic a human well enough that another human would not be able to tell. Today the test has only being passed once, just.
Hollywood and science-fiction writers have created many popular stories that deal with a future in which robots become not only as smart as humans but smarter and enslave them.
It is more telling of humanity that we assume robots would enslave us just because the human race has tended to enslave those it can dominate.
Nevertheless, the World Economic Forum has identified the impact from advances in this field to result in a Fourth Industrial Revolution which is likely.
Credit: World Economic Forum
What is A.I.?
The simple answer is that it is the process by which a computer seeks to replicate human brain function to carry out complex operations.
The truth though is that it is a group of very challenging functions that computers are trying to match in order to do some of the functions at a level similar to us. It includes:
Cameras, microphones and other sensors allow machines to rival human perception and in many cases to surpass it. Using these machines to help us perceive the world around us is a benefit and is critical for developments like driverless cars.
This is probably the component that is most developed as machines have access to a huge store of digital information. Although as you know from doing a search on the internet, just because there is a lot of information does not mean there is a lot of reliable information.
This is the area where there have been impressive gains recently as machines use experience to learn and test the information they have. Learning is also able to be shared with other machines which accelerates the speed of learning and would allow even a basic program to achieve incredible results. Google's DeepMind relies on this type of learning.
This is far more complicated that you might think as computers are precise while our speech is varied and we don't always follow the already very convoluted rules of language. So for a machine to determine not just what you said, but what you meant is a real challenge. Assuming it is confident of your request it then needs to determine if or where to find the answer and then show it to you.
This is the big one. How to make a machine interact with its environment in such a way that it can make sense of it is very difficult. Machines can react to programmed triggers or to a defined set of circumstances but that only gets it as far as the programming allows. To challenge humans, in the ways some fear, computer logic and processing power will need to improve significantly. Even those optimistic in the field don't see all the elements above coming together for some years yet.
A.I. in action
Every time you do a Google search you make use of A.I. programming that will try to determine what you are looking for from the relatively short string of letters you supply. Given there are over 4 billion pages in the indexes, knowing which one you are looking for and posting the answer in a fraction of a second is truly impressive.
Using the digital assistants from Google to Apple's Siri to Microsoft's Cortana and, the most recent, Amazon's Alexa will accept what you are looking for simply by speaking to it and then finding what you requested or performing the action you requested.
Tesla cars along with many other manufacturers are recording the huge amount of information about your driving. Tesla specifically learns to identify potential danger scenarios and shares its learning with the rest of the cars in the network. When fully driverless cars are on the roads, their programming would not have come from humans but hundreds of hours of observing human drivers.
Google's DeepMind deserves a special mention in that, rather than teach it all the possible moves for a game and then leave it to test the best solution, it simply tasked it to try to gain a high score. By first watching what happened with the game, and then noting what changed when it made a move of its own, it worked out not only what it needed to do but kept refining the process until it could play the game better than a human.
In beating a human opponent at the very complex game of Go, it did not wait to consider the outcome of each move as there are simply too many, it used a level of reasoning similar to humans. The human opponent noted after the 1st game that the computer was at first playing safe, but changed its strategy when the human opponent reacted to it.
The jobs that robots will do
Machines are becoming very good at each of the aspects mentioned above, but still have a long way to go when a task is required that combines all of them.
So jobs that require a lot of perception but not much else, such as those of security guards, are at risk. An Oxford study to determine the likelihood that a job would be automated in the next 20 years lists a security guard as a job with a 89% chance of being replaced.
Likewise, jobs that rely heavily on a set of knowledge or repetitive communication or tasks will almost certainly be automated.
Those jobs where the tasks require a variety of reasoning and logic to be used are less likely to be automated.
You might think it unfair that machines will replace humans, but the majority of the roles that will be completely automated are likely to be roles that humans did not thrive at either.
The challenge is how to accommodate those that will be affected and that raises the challenge to disrupt the one institution that really needs a major overhaul - education. We can no longer expect our system of education to supply young people with the skills that have value in our increasingly high tech world.
When the likes of Stephen Hawking, Elon Musk and Bill Gates all weigh in on the potential risks of A.I. it would be unwise not to consider it.
It does not mean they are against the attempt to improve machine function, but warn that at some point they will begin facing moral decisions about what the machines will be allowed to do. Already the debate has began about how a driverless car should deal with a scenario when it would endanger the car occupant's life to save a group outside the car.
You don't need a smart computer to tell you that we live in interesting times.