Machine Intelligence Research Institute

Write a review Add photo


	0 reviews

Description

Update

MIRI exists to ensure that the creation of smarter-than-human intelligence has a positive impact. We provide tax receipts for BTC donations!

As Stephen Hawking and others recently wrote, “Artificial-intelligence (AI) research is now progressing rapidly... Success in creating AI would be the biggest event in human history. Unfortunately, it might also be the last, unless we learn how to avoid the risks.” Humans rule this planet not because we are the fastest or the strongest, but because we are the smartest. But “this century we will cede that crown to machines,” says Jaan Tallinn, co-founder of Skype. Unfortunately, MIRI is one of only two organizations in the world with full-time researchers devoted to the problem of ensuring good outcomes from smarter-than-human AI. Stuart Russell, co-author of Artificial Intelligence: A Modern Approach, illustrates humanity's under-investment in this issue with an analogy:

“If a superior alien civilization sent us a message saying, ‘We'll arrive in a few decades,’ would we just reply, ‘OK, call us when you get here – we'll leave the lights on’? Probably not – but this is more or less what is happening with AI.”

MIRI was founded in 2000 to address this challenge. Our sole goal is to ensure that smarter-than-human AI, when it is created, has a positive impact. In particular, we focus on the technical design challenge. How do we design an AI that is transparent to human inspection, that has a principled (rather than kludge-like) design conducive to several kinds of safety guarantees, that can be given desirable goals, and that will retain those goals even as it modifies its own code to become more and more capable? To answer these questions, we host math and AI research workshops, publish papers with collaborators from around the world, and try to persuade other researchers to join the effort. Our work has been featured in the world's leading AI textbook, in The Cambridge Handbook of Artificial Intelligence, and (forthcoming) in the world's most-read peer-reviewed computer science magazine, Communications of the ACM. We hope you'll support our efforts to address the most important challenge of the 21st century.