IS ARTIFICIAL INTELLIGENCE A THREAT TO MANKIND?
It’s indisputable that the rapid advance of technology has made our lives easier. It’s streamlining jobs, speeding the delivery of information and automating repetitive tasks.
Thanks to smartphones and computers, we are the first creatures to ever outsource their own memories to machines. And thanks to smart home technologies our homes can now all but take care of themselves.
Floors, lawns and car servicing
Little robots can vacuum our floors and mow our lawns. Our car gets plugged in to an engine analyzer when we take it for service. The mechanics read the diagnosis and get to work.
Artificial intelligence (AI) can now easily beat our world champion chess player. More recently, AI beat the world go champion. Go is a much harder game to play than chess. But wait for it… the AI go champion taught itself how to play the game! It was just programmed with the basics.
And artificial intelligence is revolutionizing diagnostic techniques in medicine! For example, AI has been shown to be better at cancer diagnosis than radiographers.
The good, the bad and the dangerous
Yet despite all the advantages of developing technology and Artificial Intelligence, not everyone sees a rosy future. In fact, there are some very notable voices warning of potential dangers.
The great physicist, Stephen Hawking, saw AI as a potential apocalypse. Entrepreneur, Elon Musk, has likened the emergence of truly intelligent AI to summoning a demon. Bill Gates is a little more circumspect with his views of AI but still believes that the development of AI should be monitored and viewed with great concern.
SO WHAT EXACTLY IS AI?
While Hollywood likes to portray AI as human-like robot intelligence with nefarious intent, the truth is a little more broad.
Modern AI comes in many forms. Apple’s SIRI, Hey Google and Microsoft’s Cortana are AI. Self-driving cars and autonomous drones and weapon systems are AI. Every time you use the Bing or Google Search algorithm, you are interacting with AI.
Most of us would agree that the above examples are all beneficial. They help us get through both work and play. But what about mass surveillance systems?
Your local airport probably has a facial recognition system installed. The Chinese government has developed AI that can identify someone by the way they walk – even if they try to hide their normal gait!
Again, most of us would agree that these are examples of benign AI. After all, these are systems that are designed to protect us from the bad guys, right?
Modern AI is more correctly called Narrow AI. It’s designed to perform a narrow or limited task better than a human could.
Narrow AI poses little to no threat to humanity. Its purpose is too narrow to make it much of threat outside its limited scope.
The problems arrive with the development of General AI. While Narrow AI is capable of performing a single task much better than a human, General AI would be able to outperform a human at any given task.
AI MORALITY AND MISALIGNED GOALS
Books and movies are rife with stories of AI’s turning evil or developing a true independent conscience for some reason or other. They typically decide that to protect Earth, they need to destroy humankind. The moral imperative! But morality isn’t the real problem with General AI.
Misaligned intelligence is a much more pressing problem than notions of good and evil. If developers don’t set clear goals and guidelines for their AI machines, then the intelligence may find a solution that is expedient but potentially harmful.
For example, let’s say you are programming the AI of a self-driving car. As a part of your program you instruct the car to avoid traffic jams and always find the fastest route to a destination.
Without defining that it must obey all the road rules – run under a specific speed limit and act to avoid any and all injury, the AI may decide the easiest and fastest route is to jump the curb and speed down the footpath! Pedestrians be damned!
The AI achieved its stated objective but in an unexpected manner. Similarly, unless a specific end goal is stated, similar misalignment is possible. Again, fiction is rife with such stories and tropes. One of the more common being an automated weapons AI ending war by eliminating mankind.
I’m sure you’ve noticed that autonomous algorithms are very adept at targeting advertising at you on your social media channels. This is done by tracking and analyzing your browsing habits. These AI’s can extrapolate a huge amount of information from your social media usage – anything from age and location to interests, hobbies, health and political leaning.
In 2016, British based political consulting firm Cambridge Analytica harvested the data of 50 million Facebook users. They then used that data to try to sway the U.S. Presidential election and the U.K. Brexit vote.
Cambridge Analytica using Facebook’s AI, chopped, sliced and diced the data so that about the only thing not known was the color of your underpants.
But AI can do more than just harvest and manipulate information. It can (and has) been used to disseminate information. This has allowed propaganda or incorrect information to be funneled to those most likely to find it convincing. The Russians used this technique very successfully during the 2016 US election campaign.
DISCRIMINATION AND CONTROL
In one way or another, most of what we do is monitored. Websites gather user information and surveillance cameras watch the streets with facial recognition software.
China’s Social Credit System is a showcase of how AI can be used for control. When the system is fully operational, every one of the 1.4 billion citizens of the country will be given a score.
This score is based on trackable data about their behaviors. Do they smoke? Have they ever jaywalked? How much time do they spend online playing video games? Do they pay their bills on time?
Depending on the score, citizens may be banned from certain activities or amenities. There are already examples of citizens being refused passports and even being banned from all public transport.
ARE WE DOOMED BY AI?
Technology is amoral. It has no desire outside its purpose. It’s up to us to make sure it isn’t misused.
Narrow AI is being used for a huge number of good causes – from lymph node mapping and the potential cure for cancer to your predictive Netflix recommendation through to making your car safer.
As our AI capabilities expand, these good uses will likewise expand. But the potential for misuse or poor design could easily lead to an increase in the problems outlined above.
The march of technology can’t be stopped. It’s literally seen exponential growth since the first PC’s hit our desks 40 years ago. That phone in your hand is a thousand times more powerful than my first PC (a Commodore 64). The power is already here to drive AI forward.
At the current pace of development, some believe General Artificial Intelligence will be with us within a decade. Others argue it is an impossible dream (or nightmare) and will never be achieved.
The good news is that, whether it arrives in 10, 50 or 100 years, teams of researchers, the world over, are already working on methods of curtailing or steering General AI to ensure it works for you and not against you!
But will they be outsmarted as our chess and go champions were?
What are your thoughts on AI. Will it lead to the next renaissance or Armageddon? The jury’s out. What’s your verdict?