Artificial general intelligence – coming to a bot near you

DeepMind says it’s on the verge of a human AI.  

Dawn of the planet of the AIs 

We all know it’s been coming. Finally, one of Google’s companies thinks it’s perfected artificial general intelligence. That’s an AI that can do just about anything a human can do: Write poetry, stack blocks, hell, even bust some killer dance moves.  

DeepMind’s research director, Dr. Nando de Freitas, was even so confident as to say that “The game is over” in response to an article stating that a generalist AI is next to impossible. If you believe Dr. Nando, DeepMind’s latest prototype, Gato AI, is on the brink of achieving something big. By all accounts, Gato AI just needs to be scaled up to become an intelligent agent. Scary? A little. 

To scale up or not to scale up? 

AI has always been a bit of an ethical quagmire. What exactly should AI be allowed to learn? Who gets to decide? Once it’s been activated, is anybody allowed to deactivate it? While AI is of course a part of the tech space, the difficulty some firms are facing is just how human, or superhuman, they really want to make AI. Equip an AI with enough learning capability, and who’s to say it won’t eventually become the dominant form of life on earth. Some scientists think that’s a distinct possibility. 

To counteract the slight existential angst that AI inevitably causes, Google says it’s developed a ‘big red button’ that could shut down any AI if necessary. But with several scientists warning that some AI could already be semi-conscious, who’s really to say that the computers haven’t already worked out a way around it. Makes you think, right?