I had to write a paper on Artificial Intelligence and since I don’t like to waste my writing I figured I would also put it on my blog.
There are many different types of jobs one can do if interested in the artificial intelligence field. I will try to cover the 5 most popular careers and perhaps touch on a few others. Technically speaking, AI is a field and not yet a discipline, like programming. AI is built upon a knowledge of mathematics and numerous other sciences, in order to detect, assess and act on information. Most positions in the field require a masters or a PhD, with a strong emphasis on mathematical capabilities. That being said the most popular jobs in this area are game programmers, robotic scientists, face recognition software, search engines, and government employees in the defense industry.
In game programming you typically have to battle, compete, defend, strategize or defeat an enemy. In order to make this enemy difficult and to make the game challenging and fun it has to have some form of intelligence to attempt to outwit you. Programmers have to create algorithms to have enemies move in a way that counters your moves, or use seeking algorithms to find your position.
The average gamer has about 13 years of experience and typically plays a game because it is a challenge, so each new generation of games has to develop better AI in order to capture the attention of gamers. Most games have a general set of rules that the AI adheres to and sticks to a certain chain of events or a script, so there isn’t much programming needed to go off script or against the determined rules.
Most characters in a game are determined by finite states, such as idle, aware, aggressive, alert and fleeing. These states are activated by the human controlled player, and can be activated by checking for a certain variable. There is also reliance on predictive algorithms that are used in conjunction with collecting history on a human controlled player to predict the next move.
When it comes to AI characters in games they need the ability to perceive and find paths, to give them another level of intelligence. Sight is done with vectors, distance and angles, and if something is not within a certain angle the character doesn’t react to it. Sound is done in a similar way as well. Once you master sound and sight you can have enemies perform tasks such as cover, crash, and turn. And after they have those tasks you can then create an algorithm to find the best path available to get from a starting position to an end position.
When it comes to finding jobs in the gaming industry it is best to move to an area of the country that has a high concentration of these companies. There was a time when I lived in Austin, TX and noticed that they had a very large amount of these types of jobs comparable to other metropolitan areas. If ou look on LinkedIn there are about 400 opening countrywide in this field:
As I write this artificial intelligence in robots is still far away from trying to achieve the intelligence that humans have, with the ability to learn anything, ability to reason, ability to learn languages and formulate original ideas. But they have made a lot of progress and can replicate some elements of intelligent ability. The most basic robotic intelligence relies on gathering facts about a situation through sensors and human input. the computer then compares this information to stored data and decides what the information means. Then the computer runs through possible actions and decides which action will be the most successful.
Modern robotics has the capacity to learn in limited ways. They can recognize if a certain action was successful or not, and store that information for the next time they attempt a certain action Some can even learn by mimicking human actions. M.I.T.’s “Kismet” is a robot that designed to interact socially, by recognizing body language and voice inflection, and it uses lower level computers and lower level actions to mimic basic human automated actions.
The major drawback to AI in robotics is trying to mimic the billions of neurons and their connections that lead to higher level learning in humans, this complex circuitry is incomprehensible to AI developers. Mostly everything is theoretical right now and science is creating theories and testing them to see if they have figured out the secret to this higher level learning. So far we are pretty far away from the Authur C Clarke’s version of the future where robots and human are so integrated it enables humans to live hundreds of years.
When it comes to finding a job in robotics it would appear to be much easier, but is also entails more than just the artificial intelligence aspect which is hard to parse out in a job search. IN the field of robotics, there are currently over 4000 jobs you could find on LinkedIn:
Facial Recognition Intelligence
This kind of technology really fascinates me as you can have it being used with a pair of glasses and observe a persons face and immediately know their name and have access to their Facebook profile. No one would ever have to worry about memorizing people’s names anymore, imagine how this would work in sales, marketing and networking circles? You can immediately appeal to anyone’s interests like Facebook does with their advertising algorithms.
Privacy advocates, however, are going to be absolutely against software like this. Any person can now tell if someone is liberal or conservative, gay or straight, religious affiliation, what sports teams they like, etc… It could lead to more discrimination or it could bring us closer together as we realize how many things we have in common with another. Imagine sitting at a bar alone and knowing that the person next to you loves the same sports teams you do, or is a devout Christian?
The software is currently almost as efficient as humans in recognizing faces, and is only less effective by slightly more than .25%! Eventually this tech could surpass human recognition abilities. This is the software that Facebook will eventually use in tagging people in photos, that is why when you hover over a photo it magically knows who it is, and it is soon about to become even more magical!
When searching LinkedIn for this type of job I ended up finding gigs for places like salons, so I thought it best to check other web sites and noticed that on Indeed there was a mere 107 total jobs listed there:
So while the amount of jobs in this industry is very meager right now, you can bet that with its marketing capabilities that it could end up being the fastest growing part of artificial intelligence development out there.
Search Engine Intelligence
- March 2013: Google acquires DNNresearch, a neural network startup out of the University of Toronto, and gets the team refocused on expanding traditional search algorithms.
- January 2014: Google acquires DeepMind and sets up the artificial intelligence team to work directly with the Knowledge team on Google’s search algorithms. (They also, almost immediately, set up an AI ethics board — presumably, to save the human race from AI-wrought extinction.)
- September 2014: Google expanded research surrounding quantum computing by hiring John Martinis and his research team out of UCSB.
- October 2014: Google acqui-hires two teams of AI researchers from Oxford (and announces a partnership with the University) to “enable machines to better understand what users are saying to them.”
I posted the above listed link from here, just to give an example of how far and ahead Google is over anyone else in using search engine capabilities to create artificial intelligence. Rumor has it that they already have algorithms that can learn and are readjusting as we speak, hence why they have reduced the amount of announcements they make on their own algorithm discoveries. Part of that may be due to google switching to SEO’s that are geared more towards fact based accuracy over “hit” based accuracy, but unless you are in google’s top brass there really is no way to know.
Search Engine Optimization (SEO) still seems to be the place to go if you are entering the intelligence industry right now, but it is still very broad as it could range from simply being a key word specialist to being an actual algorithm developer with Google, so be wary of the results below from LinkedIn as they are likely not representative of actual AI jobs:
Atrificial Intelligence in the defense industry
The description below of quantum computing is taken from NASA’s own website here.
“The NAS facility hosts the Quantum Artificial Intelligence Laboratory, a collaborative effort among NASA, Google, and Universities Space Research Association (USRA) to explore the potential for quantum computers to tackle optimization problems that are difficult or impossible for traditional supercomputers to handle.
The laboratory houses a 512-qubit D-Wave Two™ quantum computer, installed at the NAS facility on the NASA Ames campus. The facility has been extensively retrofitted to provide isolation from noise and vibration, as well as the infrastructure required to cool the system to its near-absolute-zero operating temperature. The installation team recently completed a series of rigorous calibration and acceptance tests, and is planned to be operational in early fall 2013.”
Quantum computing is an exciting area of artificial intelligence. Theorists say that one day you can turn on your computer simply by thinking it or observing it, as quantum computing approaches further development we get further understanding of how this could actually happen.
As for government jobs in artificial intelligence they don’t seem to use social media as much as other private sector employers would so the jobs available through LinkedIn are fairly limited at just a few dozen or so.
Overall the field of artificial intelligence is an ever evolving ever developing industry in which the approaches to reaching human like intelligence is happening from many different industries and many different theories and methods.
And as AI develops and becomes smarter it becomes more likely that more jobs will be taken away just as they are created.