Project 1

The Potential Danger of Artificial Intelligence: Everything an Earthling Needs to Know

When most people think of artificial intelligence, they think of Terminator and Wall-E. These fully operational robots are capable of existing as if they were real humans, but in reality, our technology is not quite that cool yet.

While the abilities of artificial intelligence are not up to movie standards many are concerned that they could be in the very near future. Fears of robot takeovers and a world run by our own creations float around the internet and our news sources.

So someone uninformed on the subject of artificial intelligence might ask themselves, “Should I be worried about artificial intelligence? Will there be a robot apocalypse in a few years?” Unfortunately, I cannot provide a concrete answer to these questions. I can, however, lay out all the available facts and expert opinions available to the general public regarding artificial intelligence and its potential danger.

To begin, what is artificial intelligence exactly?

Artificial intelligence (AI) is intelligence as exhibited by some form of machine or computer. AI can exist in many different forms and can be used for many different purposes. The simplest of AI can decide what move is best in a game of tic-tac-toe. The most complex AI can manage human employees in a warehouse or measure gas flow and adjust an entire system to accommodate for fluctuation or even write award-winning poetry.

    Screenshot of AI poetry, 9/12/2015, via screenshot from my computer

Would you have guessed that this was written by a computer program? If not, then this computer program would have passed what is called the Turing test. The Turing test is a test where a machine attempts to exhibit intelligence that makes itself indistinguishable from an actual human. The test is simple. Take a blind test subject and have them interact with the machine. If the human test subject believes they are interacting with another human, then the test is passed. So, another definition of artificial intelligence is any machine that can pass the turing test.

In summation, artificial intelligence is the ability for some machine or computer to think independently and perform an action based on what it determines is optimal for the given situation.

An example of AI that many of us use everyday is our great friend, Siri. Not only does Siri take our inputs and come up with the appropriate response, but Siri also evolves and adapts based on each user’s tendencies when using her.

Here’s an interesting thought experiment. If Siri were given a more human voice and set up for a Turing test would you believe she was a real human? It’s hard to deny Siri’s irresistible charm and great sense of humor.

So you might be wondering, “aside from Siri, what other forms does today’s AI take?”. And the answer is pretty incredible.

What is AI capable of right now?

While the state of AI development has not reached levels illustrated by science fiction movies, AI is still capable of performing incredible acts.

First and foremost, artificial intelligence is capable of logical thinking. A few types of logic the AI can exhibit are fuzzy logic and propositional logic. Fuzzy logic (a value based logic system) has been used to predict earthquakes, improve the efficiency of high-speed trains, and adjust gas flow mechanisms. Propositional logic is slighlty different, and is based on taking premises and drawing logical conclusions from these premises. For example, if our two premises are “if it’s raining it’s cloudy” and “it’s raining”, then our conclusion is “it’s cloudy”.

An example of these logic systems lies in the technology that allows the disabled to speak. Stephen Hawking’s inability to speak is remedied by a computer system that senses Hawking’s cheek movements and utilizes a constantly evolving word prediction system. This predictive word system is an example of AI because it performs an action (in this case selecting words) based on what it logically deduces as optimal.

Neural networks, an up and coming branch of AI, describe machine systems that function similarly to a real brain, learning and constantly evolving. These systems take several different “neuron like components” that exchange information between each other and come up with an appropriate action or output. Below is a link to an example of a neural network where a computer learns to perfectly play a level of Super Mario.

Screenshot from the neural network video, 9/12/2015, via my computer

While a computer can now play Super Mario better than any human, perhaps what is more important is what AI is not yet capable of.

What is AI not yet capable of?

AI has yet to become true artificial general intelligence (AGI). This is the type of AI displayed in fictitious novels and books, and it is defined as the ability of a machine to exhibit complex, human-like decision making and cognizance. The distinction between artificial general intelligence and neural networks is the level of cognizance in the machine. A machine exhibiting AGI has a much deeper understanding of its surroundings as compared to a machine running on a neural network. AGI is yet to be achieved by any man-made machine, mainly because we have no idea how to create such technology.

Microsoft co-founder Paul Allen doesn’t believe this type of intelligence will occur in this century unless there are "unforeseeable and fundamentally unpredictable breakthroughs" and we stumble upon a "scientifically deep understanding of cognition". As a co-founder of one the greatest technology companies of all-time (and a current industry leader in artificial intelligence) , Allen’s opinion is hard to ignore.

Writing in The Guardian, roboticist Alan Winfield claims “the gulf between modern computing and human-level artificial intelligence is as wide as the gulf between current space flight and practical faster than light spaceflight.” Winfield is far less proven in the technical world than Paul Allen, but the roboticist’s input still carries weight.

Even though experts believe sci-fi artificial intelligence is far from reality, there’s a general sense of worry about a robot-driven future.

Who is weary of artificial intelligence and why?

A large majority of those fearful of artificial intelligence are average people with little to no knowledge of artificial intelligence and the extent to which it is developed. Many of these people have simply never been exposed to the realities of this extremely technical field. Much of the fear this group has stems from unfamiliarity with technology and what technology is truly capable of. This type of fear has been labeled “tech paranoia” and does not carry weight considering it is not based on facts.

Another key factor contributing to the AI hysteria is social media. Blogs, articles, tweets, videos, and all other forms of social media are often created to attract the most attention possible. And few things attract more attention than a terrifying headline such as “AI is as Dangerous as Nuclear Weapons”.

Twitter is a popular outlet for expressing fears towards artificial intelligence. Adrian Shaughnessy, like many others on twitter, is not a computer scientist. Adrian is, however, convinced our machines are “already beyond control”. It’s difficult to give any credit to opinions such as Adrian’s, considering Adrian is a journalist and has no professional background related to artificial intelligence.

Screenshot of fearful tweet, 9/18/2015, via my computer

Although most of the skeptics of artificial intelligence have limited knowledge on the subject, there do remain a significant amount of respected scientists, software developers, and even AI developers who are fearful of what advanced AI may do to our society.

Stephen Hawking, a world-renowned theoretical physicist, has gone so far as to say in a video interview with BBC ,“...I think the development of full artificial intelligence could spell the end of the human race.” The way Hawking phrases his concern, saying AI could literally threaten our entire existence, aims to create a deep impact on his audience. Hawking’s words appear to be deliberately chosen to alarm and create a sense of urgency and awareness in those unfamiliar with AI. Hawking’s concern with our species’s well-being carries significant weight as a greatly respected and renowned scientist.

Other respected scientists also share Hawking’s concern.

Elon Musk, CEO of SpaceX and Tesla, stated in an interview “[I] think that there should be some regulatory oversight, maybe at the national and international level, just to make sure that we don’t do something very foolish.” Musk’s concern is very practical in nature; he has donated millions to the development of AI that is safe. His words seem to be chosen carefully so that his opinion is clear, but he does not come off as arrogant. Musk is a rather credible source on the matter of AI, considering he sees the issue from both sides. He values the potential in AI, while still expressing concern for its dangers.

While many members of the scientific community are skeptical of artificial intelligence’s place in our world, many others disagree.

What is this great potential in AI? So far, it sounds like a lot of danger.

Mustafa Suleyman, head of applied AI at AI company DeepMind says that, “The way we think about AI is that it’s going to be a hugely powerful tool that we control and that we direct, whose capabilities we limit...We’re building [AI] to empower humanity and not to destroy us.” Suleyman’s strategically uses the phrase “empower humanity” to bring light to all the potential good that AI can produce. Suleyman, and many members of the scientific community believe that AI will produce far more benefits than drawbacks.

Russ Altman, a professor of bioengineering, medicine, and computer science at Stanford, said in a speech at a robotics convention “Artificial intelligence (AI) has astounding potential to accelerate scientific discovery in biology and medicine, and to transform health care.” Altman’s words are aimed at convincing his audience of the benefits of AI. Altman’s belief in the potential of AI is significant considering his status at one of the top universities in the world.

At this point in time, it is difficult to ascertain exactly how dangerous AI could be. Some members of the scientific community are optimistic about AI, others believe it could ruin us. It may be difficult to take a side.

In conclusion, should I be afraid of artificial intelligence?

While it is difficult to say whether or not artificial intelligence will eventually reach dangerous levels of development, it is safe to say we have little to worry about for many years to come. The most harm AI can do to you now is write superior poetry or beat you at chess.

However, it is imperative to stay updated on technology’s most current developments and progress. It is our civic duty to stay informed on topics that will influence the world around us. Stay up to date. Follow the progress of artificial intelligence in journals such as IBM Journal of Research and Development and IEEE Transactions on Information Theory, as well as in online articles and videos.

So in conclusion, unless a completely unexpected breakthrough occurs in our pursuit of artificial general intelligence, we have no reason to look out for threats like Arnold Schwarzenegger in a robot suit shouting "hasta la vista, baby". Even though that would be kind of cool.