Artificial Intelligence has been surreptitiously making itself comfortable in the public’s collective subconscious since long before Siri was politely reminding you whose birthday is coming up or struggling to understand any accent other than broad Californian (or perhaps that’s just my axe to grind). With origins dating back to the 1940’s, intelligent computers began with basic decoding and algebra, developing greater processing power with frightening speed in the latter part of the 20th century. By the time they came into the public eye, they were already assisting humans in firing rockets at the moon, and also, very nearly, each other.
The Threat from Artificial Intelligence
Artificial Intelligence is a complex topic, the threat it poses more so. I recently enlisted the support of the brilliant Fred Popowich, Head of the Simon Fraser University ‘Big Data’ Program. Over coffee, he explained to me that there is “a difference between Artificial Intelligence and artificial skill. A machine with artificial skill (something that has been programmed to do a single thing well) could not pose a threat in and of itself, but it could be wielded by someone, thus creating a threat. Artificial Intelligence (a machine that can learn/assess a situation and make a decision by itself), on the other hand, could pose a perceptible stand-alone threat, as it could work independently of human interaction.”
Put simply, there are two forms of AI and both pose a threat. Artificial skill is a machine that has been taught a series of processes to carry out, that can be anything from an automatic coffee maker to trajectory system on a nuclear weapon. These machines are powerful, but cannot work on their own; they are an extension of the human threat to survival, as they must be operated or “wielded” by a human.
This doesn’t mean they aren’t a threat. The recent death of a driver in Florida, who’s car’s autopilot accidentally rear-ended a truck while driving, was heralded by many as the first death by the hands of Artificial Intelligence. In reality, it was another in the long list of accidental deaths caused by artificial skill. For example, a pacemaker is an artificially skilled machine. It is programmed to watch patterns in the user’s heartbeat and respond by shocking the heart to keep it ticking as it should. When they first became widely available in the 1970’s, a study in Oregon USA found 10% of pacemakers failed within the first month. Unfortunately, it’s unlikely that the wearers of these pacemakers could run to a hospital in time. You cannot put either of these deaths to AI as there was no malicious intent, it just failed the task humans programmed it to do.
An Artificial Intelligence, on the other hand, can sum up a situation by itself and dictate the best course of action based on what it has learned from prior events similar, or from a database vaster than human knowledge. For example, it has often been said, should an AI be put in charge of Earth it would be able to use ‘Perfect Recall’ from all of Earth’s history to find the best solution to each and every issue it would come across. Fred Popowich explained, “These machines have become bigger and more powerful in the last century as computing power has become more powerful and more accessible to more people.” More people means more data, and more data means more experience these machines can draw ideas from. As such, they pose a greater threat now than they ever have.
AI taking control of the Earth has been discussed since 1921 – often in a negative manner. If the directive of the computer was to ensure the longevity of the human race, our survival would be assured, right? Well, surely the first step an AI would take if given the reigns of Earth would be to slaughter at least one tenth of the human population. Working from logic, it would see how many of the worlds issues stem from a lack of resources to go round and conclude that our best hope for survival would be a global preening of the human population.
Another benefit AI has over humans, which increases their threat to us is, bias; or more, their entire lack of it. An artificially skilled machine can program to show favor to one thing or another, or one group of humans or another. A powerful Artificial Intelligence would know the best way to submit entirely unto logic would be to ignore any bias or preference to humans. With that in mind, true AI would pose a very real threat. It would reason that the best way to ensure human survival is to ensure the environment it lives in stays the same and the best way to ensure that is to remove humans.
The Human Threat
Unfortunately, a much more time worn and substantial threat to our species’ continued survival is exactly that: our species. Long before documented history, humans have been murdering one another with a reckless abandon for literally any reason. In modern times, we have developed slightly beyond that (humor me), but these developments have led to a new kind of threat: inadvertent annihilation due to human error.
The human race has grown exponentially over the last 200 years at a rate we have never seen before. This has led to some amazing advances in communication, which has, in turn, allowed us to stop being quite so violent to each other all the time. A conundrum which could only be solved by war previously can now be discussed by a great deal more people. Ultimately, more minds are available to think of a solution that causes damage to the least amount people.
The Long Shot
The main threat to humanity, from humanity, is our ineptitude at handling our creations. Leading the charge for the threat we pose to our continued survival is the Chatham House report ‘Too Close For Comfort’, a harrowing 30 page read on how often nuclear weapons of mass destruction have almost been accidentally deployed. Of the thirteen listed near nuclear deployments, six can be directly attributed to human error. This is a perfect example of how humans not being able to control their creations could have led to the ultimate destruction of humanity (and everything else for that matter).
The second threat to consider is that of climate change. Much has been written on this topic; needless to say, I will not do the legions of scientists and their decades of research justice in an article such as this. What I can say is that the inconvenient truth of it is that we are raising global temperatures, which are destroying ecosystems, which is, in turn, a threat to our continued survival. In essence, it’s not a threat so much as an ongoing deepening issue that has already surpassed being a threat, as it’s actually happening.
The last word on human incompetence causing our ultimate downfall is overpopulation. As I’ve said, our population has seen exponential growth in the last 200 years following the Industrial Revolution, and the advances in health care it provided have allowed us to live longer and reproduce more safely. Overpopulation is a threat as we are rapidly outgrowing the resources that can support us. Once we’ve grown to a certain point, we won’t be able to sustain ourselves any longer.
It’s easy to see AI as a scapegoat for the eventual downfall of society, but ultimately (at least for the near future) the computers we are creating to be intelligent are nowhere near capable of this kind of power. They have the processing power but lack the ability to get themselves into positions from which real damage can be achieved. It’s no secret US Missile Bases still use 15-inch floppy discs in their computers, largely to keep their weapons off the grid from where they might be at risk (lessons learned from the 80’s classic ‘War Games’, no doubt).
With the points put across in this article in mind, the threat Artificial Intelligence poses to human survival is only an extension of the threat humanity poses to itself.