Secret Sauce Behind CBS’s Top Rated TV Show Scorpion
By Nitish Kannan - January 14, 2017
As my readers know I’m the biggest fan of AI, and disruptive technologies that can have meaningful impact. I’ve been a recent fan of Walter O’Brien who’s the founder of Scorpion computer services and Executive producer of the CBS hit TV show Scorpion.
Having had the opportunity to meet and interview Walter O’Brien once before, and having discussed his secret sauce ScenGen. ScenGen stands for Scenario Generator and when Walter presented me the exclusive opportunity to demo it to me and show the ScenGen working in real time, demonstrating its capabilities I could not turn down the opportunity.
Scorpion’s latest and most profound AI innovation without a doubt is ScenGen and its applications are unlimited in potential and is probably the most profound technological achievement of Walter O’Brien and even joked with me he personally uses for making rational daily decisions in every aspect of his life.
According to Walter O’Brien, “ScenGen has broad applicability. The technology has been proven to scale in some of the world’s largest companies within the most complex of environments to eliminate issues with new software releases that would otherwise cause damage, downtime or misinformation such as memory exceptions, memory leaks, crashes, failed installations, etc.”
According to Walter O’Brien ScenGen has numerous applications including the Aerospace and defense industry, financial, medical and utility services. Walter has run ScenGen across a broad spectrum of scenarious to save lives and companies, including mititgation of trillions of dollars of risk. ScenGen runs 1.6 million test scenarios per second or 250 hours of man work every 90 minutes. A technology on this scale helped identify the Boston Bombers.
Walter, shared with me over our hour and half Skype interview some background and context on the reason ScenGen was such an important product he’s developed. Mainly, because his clients like the military, DOD, banks and software companies building mission critical software need 5 nines of 99.999% up-time and there is no tolerance for error. Imagine a company like Microsoft building software. By the time a typical piece of software is developed at each stage of development, there is a 3% compounding error that adds up. Thus, by the time it gets into production through 4 levels of development there may be a 12% error. This leads companies to have massive staff to check for errors at each level and worse off, often times most companies can’t even catch the errors at each level as accurate as an AI system like ScenGen could.
So, where does ScenGen exactly fit in here? Well ScenGen turns out to be of great value as it can be applied to almost any field to think of all possibilities for a given problem. Often times a human may not even be able to think of every possibility. Worse, humans can’t even do it fast enough.
Walter described ScenGen to me in simple terms to help me understand the theory behind how the system actually works. He said, “imagine two chess computers that were left to play each other for a very long time. Well, naturally they would eventually play every possible game of chess.
So, what are the technology applications of the ScenGen software?
Well Walter went over a dozen different potential applications with me for the various applications of Scengen, however I’ll highlight a few that really stand out to help demonstrate its capabilities as an AI program.
First take a finance manager. A money manager could in theory be assessed against the best and worst versions of themselves. ScenGen operates as an AI program that can backtrack and see all possible trades they could have made and compare to the trades they did make. As you can see ScenGen applied to major trading corporations and money management services can replace workers and have a trillions of dollars of value.
Next, another application in which ScenGen is remarkable is mission critical operations. These include AirTraffic control, mission control, credit card companies as examples. Downtime for a credit card company like American Express could mean billions of dollars. ScenGen could be used to figure out all the possible ways a system could be hacked or taken down and then utilized to figure out all the scenarios of getting hacked. This is something no set of humans could even exhaustively think of.
What is truly remarkable about ScenGen is the unlimited number of applications it can be applied to. In fact there are so many that I personally wanted to highlight a few of my favorites.
As a former scientist myself, one particular application that ScenGen if applied to at scale could be useful is in drug discovery or chemical bonding and discovering drug conflicts and interactions. For instance, once the rules for drugs are taught to the AI system, ScenGen could figure out all the dangerous drug interactions that no group of pharmaceutical scientists could ever discover on their own in a reasonable period of time.
Another interesting application Walter told me about using ScenGen is for contracts and legal work. Ironically, legal work uses no intelligence today, yes my stance on lawyers are they’re overpaid. However, one day using ScenGen could dramatically help lower legal costs and make legal contracts more ironclad. For instance if you’re doing a domestic partnership or marriage contract or prenuptial agreement, running the contract through ScenGen, could help find any potential loopholes you might have missed. ScenGen could run all the scenarios you may not even think about which could lead to very expensive litigation later.
Finally, I wanted to discuss the technical aspect of the ScenGen system. ScenGen you might imagine runs in a very expensive office building with floors of servers but in reality, ScenGen is just a software program. The speed of how fast it runs depends on the hardware it utilizes, so in theory you could run a sophisticated system like AWS(Amazon Web services) or on Google’s 30 Petabytes of data centers for large scale supercomputer type applications very quickly. However, according to Walter, ScenGen could run on an average laptop on a 1Ghz system to generate 250 human hours of work or 1.5 million scenarios per second. I got a demo of this running to crack my password on my laptop as a demo. It was quite amazing and left no doubt this software is a real operational system that has an unlimited number of applications.
Having being so amazed with the ScenGen system I asked Walter, why its not used by companies like Google or Microsoft and rather it tends to be used by the DOD, large banks and mission critical operations. He mentioned it all has to do with money. The big M word money. Microsoft has a huge set of code that’s got errors compounded to it and tons of employees. They can afford the errors as Microsoft doesn’t make mission critical software. I’m currently writing this on Google Docs as I find office to be severely unreliable and error prone on my devices. I think if they had used a system like ScenGen early on they would have avoided the compounded errors their software has.
However, there is almost no room for errors at companies like American express as each day they process billions in transactions a day. Even on hour of downtime or errors in their code could be worth billions in market cap valuation as its a massive vulnerability. Also, Walter’s company Scorpion is selective on who uses a system like ScenGen, as the software in the hands of someone like Isis or a nefarious organization could have massive negative repercussions on society. ScenGen could very easily be used by terrorists to find the most vulnerable areas to attack and be used for massive terrorist strikes. Therefore, Scorpion has to be very careful in who uses such sophisticated technology.
I really think, I am very lucky to get an exclusive peek into the most sophisticated AI system I’ve actually seen run in real time working on a demo. What I do know is ScenGen is very powerful, I only wish I could get a copy for my own use… For peaceful benevolent purposes of course 😉