While everyone is fascinated by how artificial intelligence (AI) might enhance the workplace in the near future, that fascination is also accompanied by fear. It may seem like your worst science fiction nightmare come true, but AI also comes with the potential for a very dark side. A recent press release from the Max Planck Institute for Human Development, based in Berlin, which focuses on human/machine interactions, just warned of this, while offering strategies for combating the potential of a “machine takeover.”
Using theoretical calculations, the Institute’s team has determined that AI has the potential to quickly become more intelligent than its human counterparts. If such a super intelligent AI existed and connected to the Internet, the team theorized, it would have access to all of humanity’s data. With this in hand, AI could take control of all the machines online worldwide, with the potential to destroy humanity. Even scarier? The team has determined that if this nightmare scenario were to come to fruition, humans would be unable to control it. But the same team is exploring solutions to prevent this dystopia from coming to pass. One idea is to “wall off” AI from the Internet and other technical devices so that it couldn’t access data to use for destructive purposes. The downside to that idea is that it would render AI significantly less useful. Another solution in play is to develop an algorithm that would ensure a super intelligent AI cannot harm people in any way. This is a “containment” solution. It has its short sides, too. As of right now, the team weighing these options say the containment problem is incomputable. There’s also the problem of not knowing the moment AI reaches super intelligent status. For now, it’s a race against the machines, and anyone’s guess how it will all end.
