Machine learning and cheating

Machine learning is a process that gives the ability to predict outcomes using past events. During my career, I realized that I was getting very good at predicting the probability of outcomes and/or root cause of issues. My boss always referred to it as my “insight”.

After giving it some thought, I realized that observing and modelling my work environment provided me with those skills. The problem is that it’s a slow process and done at a human speed and frequency of events.

Machine learning give this kind of skill a super boost and allows you to build a computer tool to perform the the same “insight” I had, but much faster, much much faster. If I took every events I was confronted by, the strategy I took to address it and the outcome, I could build a pretty good information technology prediction tool. The inputs is called “training data”, and once the model has been trained, it’s called your “data set”. When your model is accurately predicting outcomes, you turn off its learning capability and lock in place what has been learned. That’s what an artificial intelligence neural network is all about.

Mark Rober has an excellent science youtube channel and did a pretty good video explaining the basics of a learning machine. It’s work watching if you are beginning in the AI field.

Machine learning might very well be the biggest thing to come to our daily activities since the invention of electricity. When you think about it, its limits are only limited by our imagination, for now. In the example, AI was used to cheat. That’s defies the spirit of the game, but is typical of human nature.

To counteract this tactic, MLS (Major League Baseball) approved the development of a more contemporary solution, which is to “transmit” baseball signs using Radio Frequencies instead of line of sight signals.

You can read about this in this article:

Cheating involving computers has been around for years. I remember getting frustrated at some games because other players used AIM BOTS, which virtually gives them aiming precision superpowers. I, for one, don’t cheat at games, so I I quickly lost my interest in those games. What’s the point of playing if everyone else is cheating?

Not all cheating involves Artificial Intelligence. Someone finding a loophole in the game and exploiting it seems acceptable to some. Casino cheating sometime involved computers to track probabilities of up coming cards and influenced bets. Casino’s counteraction was to put more decks in the play, and have cheating experts pay attention to winning tables.

But how does AI help? AI can do what normal humans can’t. Process huge amounts of data in record time and help make decisions. Sometime it can provide the best decision out of million or billions of possibilities. A human cannot do this in seconds. That’s why AI will most probably see incredible development in the “cheating” business first, before it finds its way into our daily lives. Science, medicine, manufacturing, are some the many fields which can benefit from using AI.

For years now, you could replace a contract lawyer with a piece of AI software which review contracts and suggest the appropriate corrections. Is this really replacing lawyers, or simply providing lawyers with tools to make their work easier, faster and accurate? You decide.

You must have heard the expression “Cheating is allowed as long as you don’t get caught.”, or “Crime pays, until you get caught”. AI can help criminals achieve more daring heists and “not get caught” and that is a real problem. What if AI could predict how stocks will behave in the market, and give cheater a huge advantage? I’m sure the cheaters have looked into this by now. Someone, somewhere will eventually use AI to give them the slight advantage over the rest of us.

But how can someone counteract cheaters that are using AI ? First, you pay attention, look at the details, find patterns, scrutinize behaviour, look for signs. Cheaters might be on their guards at first, but they are typically lazy and will become sloppy. That’s when you can pick-up on telltale signs that tells you something is wrong.

Apply a “zero-trust” methodology to everything, which means, don’t leave gaps or dead zones in your business which can be exploited. Validate and authenticate every actions, reactions and keep logs for post event analysis (Game Replay). AI is used in CyberSecurity to go thru billions of activity logs and detect and report patterns. What would take a human hundreds of years to process, can be done in minutes using computers.

Remember the Volkswagen(VW) scandal where computers we programmed to cheat on emission tests? Was the the company run by criminals? Probably not. They were just greedy germans who preferred to take the easy road to increase profits. I don’t thing AI was behind this fraud, but rather, some creative and cleaver engineering. However, if properly trained AI can help clever engineers become better at cheating. Everybody cheats given the right circonstances and facing rewards and/or consequences. Cheaters will eventually cheat on their cheating and get caught.

Finally, there is an old saying “Trust, but verify”, which means let the events unfold, but record everything and go back to perform spot-checks. This can be applied at everything, specially with information technology using a SIEM equipped with AI. All the activities on a network can be logged and processed by an AI algorithm which is an expert at finding complex criminal, or hacker, activities.

For the better or worst, AI will see incredible innovations in the near future and virtually none of this is regulated at the moment. Using AI to create is just one example. If an AI algorithm creates something, a copyright or patent cannot be attributed to that AI. This might be true in some countries, but in China, the AI “operator” is awarded the copyright/patent! Copyright & patent lawyers are about to get very very busy.

Here are few more videos I found which give proper description of machine learning and how it is used to cheat.

And Finally, don’t panic, artificial intelligence will certainly change out daily lives, but will not take over the world and become this evil overlord prophecized by some medias. This video gives compelly arguments.

This entry was posted in A.I., Articles and tagged , , . Bookmark the permalink.