The age of the learning robot is upon us. Computer scientists from the University of Washington and the Allen Institute for Artificial Intelligence report that “Skynet” like beings, self learning computers, are now available that teaches “everything there is to know about a visual concept,” according to a report on the topic.
how does the self learning computer program operate?
The self learning computer program, called Learning Everything about Anything (Levan), uses a robotic approach to search millions of books and images on the web, absorbing knowledge as an alien invader might. The goal is to teach it everything there is to know.
Self learning computer programs are also used in algorithmic high speed trading, as was recently pointed out Columbia Law blog with featured a study finding automated trading systems utilizing self learning computers have a different supervisory requirement and can escape criminal punishment.
“Causes of action that require proof of reckless or intentional conduct are unhelpful where the humans involved with the trading lack the required mental state because an ATS perpetrated the improper acts,” wrote Gregory Scopino, an Adjunct Professor of Law at Cornell Law School and Special Counsel with the U.S. Commodity Futures Trading Commission (CFTC). “As a result, improper trading practices such as banging the close (influencing the price of a futures contract by submitting a high number of trades in the closing period), spoofing (placing orders for trades and then immediately cancelling those orders before execution), and wash trading (self-dealing, or taking both sides of a trade) could be considered illegal when committed by humans, but not when committed independently by ATSs, despite the fact that the behavior would have the same effect on the markets, regardless of whether the perpetrator was a human or an ATS.” In the spectrum of high frequency trading, computers get away with murder for what mid level humans would be punished for.
ATSs have become self learning
Scopino noted that many ATSs have become “self-learning” in that they constantly modify and improve their trading strategies independently from human direction.
In other words, the self learning computers are amongst us and growing in power – but they receive preferential legal treatment. A primary legal argument in high frequency trading is that, because they are computers, a different legal standard applies to their actions.
Scopino notes changes are being proposed to address the problem. “Congress could amend the CEA (Commodity Exchange Act) to add an ATS supervisory requirement to the CEA… in circumstances where ATSs manipulated the prices of futures contract or engaged in other disruptive trading practices that harmed other market participants.” In other words, when a computer commits a crime that would normally be punished under human law, the duel standard of justice would be eliminated.