Warning: Trying to access array offset on value of type bool in /home/hamiltw23/public_html/wp-content/plugins/slideshare/slideshare.php on line 162
Warning: Trying to access array offset on value of type bool in /home/hamiltw23/public_html/wp-content/plugins/slideshare/slideshare.php on line 165
Artificial Intelligence
Once a figment of Hollywood’s wildest imagination, the world moves closer and closer everyday to developing the types of artificial intelligence (commonly referred to as AI) you see in movies. You know, those hyper-intelligent supercomputers that want to wipe out humanity, the robot running around holding a sawed off shotgun with half of his plastic “skin” coming off his face (oddly looking way too much like Arnold Schwarzenegger), or the humanoid sex robot that who treats “wishes” as if they were “commands”. Though these examples are clearly over dramatized, there are plenty very real and functioning AI’s that exist today. Humanity has only just discovered the tip of the iceberg in terms of our knowledge of AI, and we are learning and developing new and better AI systems every day. This article is meant to not only educate you on how we have arrived at where we are today, but discuss both the positives and negative externalities of where we may take AI in the future.
What is Artificial Intelligence?
Before jumping head first into the deep end, it is necessary to establish a running definition of what artificial intelligence actually is. AI refers to “any device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals”(Clocksin). More specifically, artificial intelligence is an umbrella term that encapsulates four distinct groups: reactive machines, limited memory, theory of mind, and self-awareness.
-
- Reactive Machines: This classification refers to AI in its most basic form. These systems are purely reactionary, the have neither the ability to form memories nor to use past experiences to inform current decisions. For the best example of a reactive machine, look no further than IBM’s Deep Blue (a program designed to solely to play chess) which defeated the reigning world chess champion in 1997. Deep Blue identifies the pieces on a chess board and is aware how each moves. It can make accurate predictions about what its opponents move might be next. It can choose the most optimal moves from among the possibilities. But, once the game is over it is like it never happened. Deep Blue has no concept of the past, nor any memory of what has happened before, thus the system exists only in the present.
- Limited Memory: This is the second classification of AI, which extends the systems’ abilities a bit farther than relative machines. These systems are able to look into the past. A handful of the self-driving cars the exist today have this ability. They are able observe and record the movement of the cars’ around them. Rather than just taking a snapshot of their environment at one period of time, the car must identifying specific objects and monitoring them over time. This entails that the cars have the ability to retain information from the past (relative to the condition of the environment they are in). The cars then take this information and combine it to a data bank of preprogrammed models of the world. These models include lane markings, traffic lights and other important elements, like curves in the road. But, theses cars are unable to store this data and learn from it like human drivers can.
- Theory of Mind: This is the line that the AI systems of today have yet to cross. This stage is derived from a theory in phycology- “The Theory of the Mind” – which states that the understanding that people, creatures, and objects in the world can have thoughts and emotions that affect their own behavior. Essentially, AI systems that would be included in this category would be able to perceive that the entities around them have thoughts, feelings, and expectations for how they want to be treated. And, the system would be able to adjust its behavior accordingly.
- Self-Awareness: This is the final step of AI development: building systems that can form representations about themselves. Ultimately, the most significant challenge humanity will face in pursuing self-aware AI is that we have to not only understand consciousness, but build machines that have it (or can at least simulate it). The distinction between the theory of the mind and self-awareness is that the self-aware AI system not only will be able to perceive and react to the emotions of the environment around it, but it would be aware of itself. I would understand its internal states and emotions, and be able to predict the emotions of others based on what it “feels.” A good example of this phenomenon in humans is when you are at the airport and you hear someone exhale loudly, you are able to tell that that person is angry or frustrated, because that is how you would respond if you were frustrated or angry.
A Quick History Lesson
Now you have a deeper understanding of what AI means, lets explore how humanity happened upon the idea of artificial intelligence. Like most ideas, the origins of AI can be traced back to the ancient greeks. Below is a detailed timeline of the history of AI:
Assuming you took the time to took the time to go through the entire timeline, you will notice that AI, though the term itself is quite young, is extremely old in concept. Personally, I find it quite amazing that in the 1500’s, philosophers and writers were able to imagine something that is becoming a reality today. It makes me wonder, how many of the things we deem impossible today, will become a reality in the future? Unfortunately, the things that we have imaged for the the future of AI haven’t always looked great for the future of humanity.
Examining AI Through American’s Favorite Pastime: Movies
The vast imaginations of movie screen play writers never ceases to amaze me, especially when it comes to science fiction. Not only do the films they create offer insight into what the future potentially will look like, but they are a good reference for how the public perceives the introduction of a new technology. Historically, movies can be both trend setters and trend followers: if a movie shows new technology before the public has already developed an idea about the deemed in as dangerous or beneficial, then it is up to that movie’s depiction of the technology on how the masses will view it; likewise, if the general public has already deemed a new technology as positive or negative, then the movie will depict that technology in a way that sides with the public. This “ebb and flow” of movies being both trend setters and trend followers is a great lens in which to view the growth of AI in the eyes of the public since its debut on the big screen in 1927. Below is a relatively short and detailed presentation on the effect movies have had on the perception of AI:
What Roll Does AI Play in Our Daily Lives?
Believe it or not, you most likely intact with some form of AI every day. Artificial Intelligence is constantly working to make your life easier; wether you are checking emails, getting driving directions, browsing music, or searching for a good movie to watch, AI systems are running to make the process more convenient.
Social media has become an increasingly integral part of our daily lives. Most all the common forms of social media employ AI in one way or another. Here are a few ways popular forms of social media use AI:
Instagram: Instagram uses a combination of big data (literally just a large volume of data available to the company) and artificial intelligence to offer the user targeted advertising, combat cyberbullying, and remove offensive comments. As the amount of content on the platform grows, the development of faster and more accurate systems of artificial intelligence is critical to be able give users the information they will likely enjoy, fight spam, and enhance the their experiences.
Twitter: From personalized tweet recommendations (ie both profiles and individual tweets), to fighting inappropriate or racist content, Twitter has begun to use artificial intelligence behind the scenes to enhance their product. They process lots of data through deep neural networks (a form of AI) to learn over time what users preferences are and match them to the content that best suits their online profile.
Facebook: The grandfather of social media, will it ever die off? Well, with its institution of new AI systems, it probably won’t be soon. The AI program deep learning aids Facebook in drawing value from a larger portion of its unstructured datasets (almost 2 billion people updating their statuses 293,000 times per minute) to personalize its users facebook status.
I’m sure the perceptive readers out there already picked up on this, but ease of access and personalized interfaces do not come with out a price. In providing the most interactive and personalized experiences for users, the AI programs must have a base data set of each users personalized preferences. Platforms store data on what you have liked, people you have searched, time spent on posts, as well as many more indicators of preference. The AI then uses this data and builds a profile for you, the user, and then filters your feed ensuring that the majority of see what you see caters to what you have previously liked or retweeted.
There are also many places were AI is used subliminally, it is so minute that you might not even notice, but if it wasn’t there I’m sure you would. When you begin to type in Google’s search bar, and if begins to predict the rest of what you are typing, thats AI; When you casually ask Siri (or Cortana, or another virtual assistant) what the time is, that is AI; when you misspell a word while texting, and the phone auto corrects it, thats AI; when you take a flight across the country, the pilots of the plane, on average, spend roughly seven minutes actually flying the plane, the rest is done by AI; when your email automatically moves the ad for the new “Toothbrush that doubles as a buttplug” from your inbox to spam, (I’m sure you guessed it) it’s AI.
My point is that AI exists everywhere, and it is very important to acknowledge its presence. The first step in preventing a issue from arising, is knowing where it may come from. It is foolish to assume that AI won’t become an even larger part of our lives in the future, because it will. We must have a collective understanding as a species that AI has the potential to either be the best or worst thing to ever happen to humanity.
Potential Futures With Artificial Intelligence (Short-Run)
It is unlikely (but not impossible) that any of the generations that are alive today will live to see an AI system that passes into “The Mind Theory Stage,” let alone become “self-aware.” That is not to say that those of us alive today won’t see drastic improvements in our quality of life from AI. In the short-run (roughly 100 years into the future), there will likely be artificial intelligence programs that will be able to diagnose diseases much faster and more accurately, self-driving cars that limit the number of vehicle related deaths, AI piloted spacecraft that vastly expand out knowledge of the universe, and/or AI systems that counter humanity’s detrimental impact on society. All the AI technology I have previously mentioned is being developed at this very moment. More likely than not, many great things will come from AI relatively soon.
Likewise, with every good thing AI does in the short-run, there is a bad. Automation of the shipping industry will cause significant job loss across the country, radiologists will eventually become obsolete (AI will be able to perform radiographs faster and more accurately), any many more unforeseen issues will arise with the continued growth of AI.
Potential Futures With Artificial Intelligence (Long-Run)
Unfortunately, in the long-run, the stakes become even more dire. This idea is a bit hard to understand, but if you loose those pesky constraints caused by the way you would typically view time, then this will begin to make a lot more sense. Essentially, the idea is that artificial intelligence will be last invention man will ever have to make. In order to understand this you must only make two assumptions: 1. that humans will survive long enough to create an all-knowing self-aware AI and 2. we have the capability of creating an ultra-intelligent AI. Whether or not you can accept these two assumptions is significantly tied to the way you view time, in particular assumption 1.
To the best of our knowledge, the universe to be roughly 14 billion years old. Based of fossil records, the modern human (Homosapien) has only existed on Earth for about 200,000 years. In those 200,000 years, lacking the vastly improved technology we have today, the human race survived. If you can accept that the human race will at least make it another 200,000 more years, then it likely that we will have the an ample amount of time to create this ultra-intelligent AI (assumption 1). The second assumption is a bit trickier; it is almost purely a personal judgement call on whether you choose to accept the idea that humans are capable of developing a ultra-intelligent artificial system. If you look at the strides we have made in AI research in the last 50 years, it is not out of the question, but it could just be one of the few true impossibilities that humans face. Luckily, humanity has a loop-hole: if we can just create a machine that is both capable of learning and building other, better, machines. Today, this process is know as recursive self-improvement, and remains theoretical idea. But, we are seeing potential traces of this ability in the neural nets we have already developed. A machine that has the ability recursive self-improvement, along with the programmed knowledge of how to build other machines, would be able to start a chain reaction: Machine 1 would build a better version of itself (more intelligent and capable), then machine 2 (which was build by machine 1) would then repeat the process, this would continue until this ultra-intelligent AI with machine building capabilities would arrive. Now knowing this, then assumption 2 is a bit less far fetched.
So, why would humanity want this “this ultra-intelligent AI” anyways? Well the theory is that if this AI existed, humanity would never need to build anything again. Whatever humans could possibly create would pale in comparison to what the AI is capable of; if any problems that might arise, the AI could fix them; and questions that humanity would have, the AI would have and answer. Humanity would be in a dystopia; the AI would solve world hunger and repair the environment, the problems of today would be a thing of the past.
Unfortunately, there are a few BIG risks along the way. The first is that on our path to creating machine 1 (the machine that would start the chain reaction leading to the ultra-intelligent AI), we would create an AI program that causes the end of humanity. This is best explained through the idea of the “Paperclip Maximizer” AI. This idea was first presented by Nick Bostrom in 2003; essentially, the idea that even an AI that was built without the intention of malice, could spell disaster for humanity. The example he uses goes as follows: We create an AI with recursive self-improvement capabilities with the sole purpose of collecting paper clips. The AI would continually improve itself over and over, just to become better at collecting paperclips. The smarter and more efficient the AI become, the more paperclips it can collect. Eventually, it will run out of paperclips to collect; when that happens, the AI will develop a way to produce more paperclips in order to collect them. Then, the AI either realizes that it is competing with humanity for resources that it needs to make paperclips to collect and decides its in its best interest to eliminate humanity; or, the AI develops a way to convert matter into resources to make more paperclips to collect and just starts turning humans into paperclips. Either way, it doesn’t end well for humanity. The second big risk is that we do develop this recursive self-improving and ultra-intelligent AI, and are unable to make it have any compassion towards humanity. From there, the AI either views humanity as a threat, and eliminates us; or, the AI would decide to leave earth, in which case humanity is back to square 1.
I hope that you were able to broaden your knowledge of the vast field of artificial intelligence through this project. Please keep in mind that nothing in the future is certain, but it is important to stay aware and vigilant, because anything is possible.
Works Cited:
“Artificial Intelligence News: Latest Advancements in AI Technology.” Business Insider,
Business Insider, https://www.businessinsider.com/artificial-intelligence.
Bolter, J. David. “Artificial Intelligence.” Daedalus, vol. 113, no. 3, 1984, pp. 1–18. JSTOR,
www.jstor.org/stable/20024925.
Clark, Josh. “Radiolab.” Radiolab, Radiolab, 16 Nov. 2018.
Clocksin, William F. “Artificial Intelligence and the Future.” Philosophical Transactions:
Mathematical, Physical and Engineering Sciences, vol. 361, no. 1809, 2003, pp. 1721–1748. JSTOR, www.jstor.org/stable/3559219.
Mihajlovic, Ilija. “How Artificial Intelligence Is Impacting Our Everyday Lives.” Medium,
Towards Data Science, 17 Nov. 2019, https://towardsdatascience.com/how-artificial-intelligence-is-impacting-our-everyday-lives-eae3b63379e1.
“Paperclip Maximizer.” Paperclip Maximizer – Lesswrongwiki,
https://wiki.lesswrong.com/wiki/Paperclip_maximizer.
Rohde, Klaus, et al. “Benefits & Risks of Artificial Intelligence.” Future of Life Institute,
https://futureoflife.org/background/benefits-risks-of-artificial-intelligence/?cn-reloaded=1
Sara. “90 Years of AI in the Movies: What’s Changed (and What Hasn’t).” Enlightened Digital, 9
Aug. 2018,
https://enlightened-digital.com/90-years-of-ai-in-the-movies-whats-changed-and-what-hasnt/.
“Understanding the Four Types of Artificial Intelligence.” Government Technology State &
Local Articles – E.Republic, https://www.govtech.com/computing/Understanding-the-Four-Types-of-Artificial-Intelligence.html.
