The thought of a true artificial intelligence scares many people out there because an AI able to think might not have the same goals and desire to protect life that most humans have. We can look to movies like the “The Terminator” for a worst-case scenario of what could happen with an AI run amuck. While a Terminator scenario might not happen, something along the lines of what happens in the TV show “Person of Interest” could be a more likely result of an out of control AI. Google’s AI team thinks they have a solution to prevent an AI from going rogue.
That solution is simply an off switch called the “big red button” says the Google DeepMind team. Researchers Laurent Orseau and Dr. Stuart Armstrong wrote in a paper that AI agents are “unlikely to behave optimally all the time.” The paper stated, “Now and then it may be necessary for a human operator to press the big red button to prevent the agent from continuing a harmful sequence of actions – harmful either for the agent or for the environment – and lead the agent into a safer situation.”
The team wants to build something called “safe interruptibility” into the system. The researchers wrote, “Safe interruptibility can be useful to take control of a robot that is misbehaving and may lead to irreversible consequences, or to take it out of a delicate situation, or even to temporarily use it to achieve a task it did not learn to perform or would not normally receive rewards for this.” This big red button might not work in all situations the scientists admit, “It is unclear if all algorithms can be easily made safely interruptible.”
Dr. Armstrong believes that it is critical that safeguards be placed on AIs before robots are able to outwit their creators and accidentally or purposefully cause harm to human operators. He believes that machines might be able to take over financial markets, transportation, healthcare, and other systems at speeds that humans can’t understand and will be capable of doing much more than a specific and limited set of tasks.