Artificial intelligence began with, as Pamela Mccorduck, a science fiction author writes, “An ancient wish to forge the gods.” Now, in the twenty-first century we are coming extremely close to realizing this dream. Within the 47 billion dollar industry there are many notable artificial intelligence development projects currently being pursued such as Google’s DeepMind and AlphaGo as well as IBM’s Watson. These AIs are only the first among what looks to be a long line of artificially intelligent robots. Before these primitive AI’s such as Siri or even more primitive one’s like the chess AI on your computer existed. It is likely that in the near future AI will become an integral part of society and you will have to interact with intelligent machines daily, but how far off from this goal are we?
How Does AI Work?
Before one can understand what artificial intelligence is.
AI is, simply put, a machine capable of taking in inputs, visual or otherwise, and uses those inputs to maximize its chance of success at some goal.
Beyond this however AI should be able to learn from past successes and failures allowing its efficiency at completing tasks to increase.
So how does a AI take in inputs?
Some are easy, the machine can simply read in data from another instrument.
Other forms of input such as visual can be much harder to recognize and can take a large amount of complex programing.
Often machines have to be literally taught how to recognize visual inputs. The way the AI researchers do this is through a process called machine learning. The first step is data collection; researchers gather thousands of points of data for each of what they are trying to teach the machine. If they were attempting to teach the machine to identify cars they would find thousands of pictures of cars. The next step is data preparation; the researchers randomize the data order, make sure that there are no duplicates or outliers in the data and then split the data into two groups; the larger group is the training group, and the smaller is the test group. Now comes the training, the AI is shown thousands of images or other data along with the meaning for that data. For the car example used earlier, the AI would be shown thousands of images of cars and the researchers would identify the car in each one. As this happened thousands, maybe millions of times the AI would begin to be able to associate certain characteristics in an image with a car. They would be able to see for example that all cars have wheels, windshields, doors, windows, and hoods. The final step of the process is the test stage, during this step researches show the AI the data from the test group in order make sure that the AI is able to use inputs to recognize the an object. Machine learning can be applied to many different things, from recognizing a car in an image to learning how to play a game. Machine learning is the backbone of artificial intelligence.
Google DeepMind's AlphaGo
AlphaGo is artificial intelligence robot developed by Google’s DeepMind division.
AlphaGo recently beat the go world champion, Lee Sedol, in 2016.
This is an extremely significant milestone for artificial intelligence due to the 2.08 x 10^170 possible moves throughout the course of a game making the traditional method of calculating every possible move for the entire game used by “AI” Chess and other game robots impossible.
Instead Google had to literally teach AlphaGo how to play the game.
The scientists working on AlphaGo were able to teach it how to play go by first having it watch 1000 games of amateurs playing go. From these games AlphaGo was able to learn very basic rules and strategy used in go. From this point AlphaGo played itself one million times. This was done so that AlphaGo could begin to develop more complex strategy and begin to improve. After these games Google had a new and more strategically advanced version of AlphaGo, already it was evident that AlphaGo was capable of learning. This new version of AlphaGo played the older version one million times. These games caused AlphaGo to continuously grow and develop. This process was repeated over and over again until AlphaGo was able to beat the earliest versions of itself 100% of the time and the previous version with an 80% success rate. This method of machine learning is extremely promising and the AlphaGo team is applying it to other things such as video games. Currently they are teaching the AI how to play classic 80’s arcade games like space invaders but they have high hopes for what it will be able to learn and do in the future.
IBM’s Watson is a AI supercomputer that combines artificial intelligence and highly advanced analytical software in order to read vast quantities of data and answer questions based on it.
In other words Watson is a fancy magic 8 ball.
One of the best things about Watson is that it can serve as a platform for other programs.
In 2016 law firm signed a deal with IBM for a legal expert system based on Watson.
The system called ROSS has the capability to read through billions of documents, analyze the information, and answer complex questions asked by the law firm’s lawyers in under 3 seconds.
A similar program is currently being developed to assist medical research.
The speed at which Watson is able to answer questions makes research and legal work for more efficient.
Watson is the perfect example of how AI’s have the ability to change the future by increasing efficiency. Its analytical software is able to locate documents pertaining to a certain question and its AI components are able to intensely comb through those sources for data pertaining that can be used to answer the question and form a coherent response using that data. The speed at which Watson can answer question will transform the speed at which we make discovery. Other AI’s like it are on the horizon and as they emerge they too will increase the speed of discovery.
What are Some Things to Keep in Mind?
Everyday in the world, new technological materials are created.
As AI machines begin to appear everywhere the safety of the future can expect to affected.
Artificial intelligence is no different.
So, on the face of it, the Safety Principle seems like a “no brainer,” as Harvard psychologist Joshua Greene described it.
It’s obviously not in anyone’s best interest for an AI product to injure its owner or anyone else.
This could hurt the future forward progress of AI and the great reputation AI has for safety from a massive capacity of data.
The only way for AI to remain safe in the future is for precautions to be taken when sending out AI machines and to take the process slowly.
A few of the possible safety concerns for AI in the future are listed below:
Destruction of Peace
Artificial Intelligence could enable robots to fight, and those Artificially Intelligent robots could be used as a weapon against other people. By governments or by terrorist groups, an army of Artificially Intelligent robots could be unleashed upon a desired enemy. These robots could be armed with state of the art weaponry, and with more advanced algorithms and sensors, they would never miss a shot. They could throw grenades with perfect accuracy, they could take down human soldiers in seconds. Wars would boil down to one simple question, who has more advanced robots?
Innocent People's Loss of Life
As already mentioned, Artificial Intelligence could be used by terrorist groups. If these groups want to send a message, they could use artificially intelligent robots as suicide bombers. The robots would be able to determine when and where they could harm the most people, and they would never second back out of the bombing. There is also the possibility of an artificial intelligence going rogue. If it wanted, a rogue artificial intelligence could kill many humans without ever being stopped.
More Advanced Terrorist Plots
With unlimited resources and extremely fast processing power, an artificial intelligence could think of ways to harm the most people and how to send a message to humanity. Terrorist groups could certainly use an artificial intelligence to help them figure out more complex plots and nightmarish ideas.
If an artificially intelligent robot where to go rogue, it could cause a mass panic.
Electricity could be shut down, waterways and plumbing could be broken down, the world wide web could be attacked.
Though the possibility of an artificial intelligence going rogue is incredibly slim, it is a possibility we must be prepared for.
Some other concerns are Destruction of Property, Electrical Shortages, and an Economy Crash.
Potential Future Uses
What is There to Look Forward To?
Artificial Intelligence could completely change how we live life and perceive our purpose.
Artificial Intelligence has endless capabilities, it could improve our lives tenfold.
This Artificial Intelligence Revolution that could be coming up could create many new jobs and free up our time.
There are many things to look forward to in the future.
A few of the potential uses for AI in the future are listed below:
Imagine if you had your own teacher who knew exactly how you would learn information best as a kid.
Think about how much better you could have performed in school if you had access to a private tutor.
Artificial intelligence could do this.
The defining characteristic of AI is the ability to learn, so an artificially intelligent teacher would be able to adapt its teaching style to fit each student separately.
This would improve student’s chance of success dramatically.
Many children, especially in the 21st century, are hyperactive and have trouble sitting still in school for extended periods of time.
This can make teacher’s jobs far more difficult because they cannot adapt the entire curriculum to fit the needs of these students.
An AI teacher however would be able to learn how to best teach each individual student.
This would allow students who do not do well with more traditional methods of teaching to have a better chance at success.
Furthermore students who do exceptionally well at school would be able to explore enrichment opportunities more frequently.
In addition to personalizing education AI would be able to easily monitor all branches of a student’s education. This would allow the AI to better coordinate large tests and assignments as to not give the student to much work at one time. Often children, high schoolers especially can be easily overwhelmed by the large amount of work that they have to do. It can be difficult to balance extracurriculars and school work especially with a number of large tests and papers all at the same time for different classes. AI would be able to better coordinate between different classes so the student would not get overloaded with work. This would decrease stress and overall increase productivity.
Research is an essential part of scientific discovery.
It leads to the generation of hypotheses and experiments.
Without research it is likely that Watson and Crick would have never discovered the double helix that forms DNA, or that Einstein would not have created the theory of relativity.
Despite its importance researching is an incredibly time consuming portion of an experiment taking time away from other important parts.
However, AI has the capability to change this by greatly improving the efficiency of research.
It, in fact, has the capability to change this by greatly improving the efficiency of research.
It, in fact, has the capability fully research any topic independent of human assistance.
Currently data mining software does exist however this software is incapable of compiling data or sifting through the documents themselves. AI can do both of those things. Researchers and scientists would have the ability to teach the robot exactly what type of information they were looking for, allowing the AI to create a much shorter summary of data. This would save an immense amount of time for researchers who could simply tell the AI what information that they were looking for and allow it to gather data. Computers, which are far more adept at gathering data than humans would take a far shorter period of time. AI would save researchers and other people looks for information days, weeks, or months.
Overall the existence of AI would completely revolutionize research as we know it making it far far faster and reliable. The speed of discovery will increase massively as a result of artificial intelligence and the rate at which new technologies are invented will increase as well. AI will change the face of science by changing the face of research.
Social interaction is at the core of what makes humans humans.
As children, and even adults one prominent form of social interaction was play.
At times however, it is often difficult to find a human companion to play with.
Artificial intelligence in the future will have the ability to act and make decisions like humans.
A prime example of this that has already happened is Google DeepMind’s AlphaGo; the robot which beat the Go world champion in 2016.
AIs like AlphaGo with the ability to make complex decisions would completely revolutionize playing games.
AI with the ability to act like a human would be the perfect partner for playing a game, whether it be a physical or video one. Most video games today are plagued with false AI practice tools; AIs that cannot really learn. This means that they are incapable of simulating humans, and therefore do not make the best decisions. This causes them to be extremely easy to play against which prevents them from fully helping players to practice. Truly artificial intelligence would be able to learn providing better practice and even large challenges to players. In other games such as chess AIs are the perfect learning tool. If both the player and the AI know little to nothing about how to play chess than they will progress at the same rate as they play each other. This would give players a companion who was able to perfectly match their skill level.
AIs recreational capabilities are somewhat limited as most sports require an amount of athletic prowess that machines are incapable of simulating. This means that they cannot play sports while acting like a human. It seems that at least with robotics technology available in the foreseeable future AI will not have the ability to play sports.
Despite the small drawback of being unable to play sports artificial intelligence is an superb recreational tool. Its ability to almost perfectly simulate human interaction means that it is able to play games as well as humans providing a bored person a friend who is always ready to play a game.
There are many ways that artificial intelligence can make life at home easier.
Already we see robot home systems such as Amazon’s Alexa to increase efficiency at home by responding to commands.
AI would be able to take this to the next level by predicting whether or not to increase the temperature or dim lights in a room.
As an AI program learned different people in a home’s preferences it would be able to adjust the temperature of a room to meet that than a certain person preferred.
One of the things that can take the most out of a day is shopping, artificial intelligence however, would be able to accomplish it easily and in little time. Using programs that allow food to be ordered online AI would be able to both plan out and by the ingredients for an healthy and delicious week of meals. With access to recipes all across the internet AI would be able to easily sort through and find those that best fit the taste of a family. By using a rating system AI would be able to adjust the meals that it planned and ordered to make sure that everyone enjoyed them.
Artificial Intelligence has the chance to have a significant impact on the commercial industries of the United States and the world.
AI is currently being used for transportation for riding sharing programs like Uber and Lyft an for autopilot in commercial flights.
While AI has been used in commercial airlines for many years, the uses are becoming more advanced, and it was recently reported that on average, pilots only steer the plane for seven minutes a flight.
Through ride sharing programs and self driving cars, AI will help limit traffic on roads and therefore transportation will be become more efficient.
Commercial uses should be expected to take over jobs in the next two decades. At grocery stores, cashiers will become robots or all self checkout lanes as Amazon is experimenting with. Shelves will be stocked with food automatically eliminating almost all jobs at stores around the world. Salespeople will be eliminated from the jobs with the availability of robots and AI to do the same exact jobs while saving the company money. The commercial uses for AI machines are limitless in this new age and the economy has the potential to flourish and possibly stumble.