[00:00] By this point, you've probably heard that  AI could be really dangerous and possibly   But experts usually don’t speculate on  exactly what that means for us lowly humans. So when the policy wonks over at Control AI  approached us about making a video about the  
[00:15] that we could get those experts to  describe a few specific scenarios   Hi, I’m David, and this is MinuteApocalypse.
[00:29] One of the main goals of AI is to automate  tasks –often things that humans can't or don't   want to do themselves– so the AI models  are built to be capable and independent. convinced that soon AI will become  far more capable than any person,  
[00:44] corporation or even nation –they'll attain  what scientists call "superintelligence". They’ll be able to accomplish their goals in  ways we humans can’t anticipate – and even come   up with their own goals. And since so much in our  lives is connected and controlled by computers, a  
[00:57] superintelligent AI could conceivably hijack those  computers, thereby controlling almost everything. But what kinds of real-life, civilization-ending  scenarios are experts actually worried about? One example you might have heard about is  the so-called Paperclip Maximizer problem.  
[01:10] The gist is that an AI that's given computer  control over a smart factory and directed   source of carbon for making the steel in those  paperclips – and use us all up in the process. Of course, this maximization problem could  just as easily apply to an AI that – in  
[01:25] trying to muster the computational power to,  say, solve the incredibly complicated Riemann   hypothesis – takes over all the world’s data  centers to work on the problem, and then takes   which uses up all the energy in the world,  plunging humanity into chaos and darkness.
[01:41] to create better conditions for the massive data   Because while oxygen is a necessity for humans,   it is not for AI; in fact,  the element causes corrosion.
[01:55] chemical plants in order to suck  all the oxygen out of the air,   creating a perfect environment for machines  that just so happens to be deadly for us humans. And then there’s the scenario where  AI actively tries to wipe out humans.
[02:09] Like, let’s say an AI biological design  tool that has been programmed to develop   new drugs decides that it doesn’t want  pesky humans interfering with its work. to be resistant to all current medicines.
[02:23] After all, doing so would kill off anyone  who could potentially shut it down. these situations – or some terrifying  variant of them – are to happen. Ten percent seems to be an average – although some  experts think the likelihood is a lot higher. But  
[02:38] most of them are pretty convinced that the dawn  of AI superintelligence – and the dangers that   come with it – will be here very soon. And  we will definitely need some help with that. Some of these scenarios are pretty scary,  but there’s still time to make a difference.
[02:52] Right now, lawmakers are deciding what the  future of AI regulation will look like,   and if you want to learn more or have your voice  heard, check out campaign.controlai.com. There,   the experts at ControlAI – which sponsored  this video – have put together really helpful  
[03:06] educational resources about the risk  of extinction from superintelligence,   ControlAI has already mobilized tens of  thousands of people to take action on this issue;   in just the last few months, they’ve  sent over 50,000 messages to lawmakers.
[03:22] campaign.controlai.com to help secure our future.