Eight years ago, a little-known researcher named Stephen Hawking predicted that artificial intelligence would either be the single greatest invention of humankind—or kick start the apocalypse.
The question remains: how can humans control an AI that has us beat in intelligence and virtually every other field?
We can’t argue with a growing number of scientists and entrepreneurs, including Tesla techno-king Elon Musk—at least not as we are now.
Welcome to Kevin’s Science Korner, a video series diving into the strange and fantastic corners of science and technology.
Let’s start with the end of humanity.
The paper clip problem: how AI might accidentally kill all humans
The “paper clip maximizer” thought experiment was first coined by Oxford philosopher Nick Bostrom in 2003. His influential theory describes how a superintelligent AI might make the very unintelligent decision to drown Earth in stationary.
According to Bostrom, a superintelligence is any intellect that “vastly outperforms the best human brains in practically every field, including scientific creativity, general wisdom, and social skills.”
This AI is so smart, it could improve itself by building better hardware than any human could, or by rewriting its own source code. The updates would come faster and faster as the AI grows smarter, eventually leading to an “intelligence explosion.” At this point, it might see humans as humans see ants, Bostrom wrote.
This could be a very good thing if we’re careful. AI could revolutionize space travel, medicine and computing. Depending on the phrasing of its programming, it could also (theoretically) kill all humans.
An AI programmed to make as many paper clips as possible might realize it could finish its mission way more efficiently without any humans getting in its way. Eventually, it might turn “all of earth and then increasing portions of space into paper clip manufacturing facilities,” Bostrom said.
Neuralink: If you can’t beat ’em, join ’em
So we learned it’s hard to stop an AI hell-bent on flooding the world with paper clips, at least with our current puny monkey brains. But what if there was a way to make us smarter?
The world’s richest man has a plan to do just that, and it involves drilling a hole through your skull and putting electrodes in your brain. Enter Neuralink, Musk’s mysterious brain-machine-interface company.
For now, Neuralink aims to let you control a computer with your mind, bypassing any need to type on a keyboard or move a mouse. But Musk’s plans are far more ambitious; he aims to eventually merge the human brain with AI.
In time, neuralink would unlock a “tertiary level” of the brain that can be linked with AI, Musk claims. This could improve our computing power and give us new abilities like being able to save and replay memories, like that one Black Mirror episode, Musk said in 2020.
Musk has long been vocal on the topic, saying in 2016 humans will end up the “house cats” of AI. In 2020, I have slammed Google’s DeepMind AI project for flying too close to the sun.
Meanwhile, some experts doubt that a “symbiosis with artificial intelligence” is even possible.
Mind uploading: could we live forever online?
Sure brain-machine-interfaces are fun and all, but we’re still trapped in these fleshy bodies. This doesn’t have to be the case, say researchers in the field of whole brain emulation.
These researchers argue that consciousness and everything else that makes us human results from the billions of neurons in your brain, the trillions of connections between them and countless the neurotransmitters passing through.
Assuming they’re right, and consciousness isn’t the result of some intangible spirit, we could theoretically make an exact model of the brain and all its connections using software. This model would then be a sentient, exact replica of the original brain.
Brain-machine-interfaces like Neuralink’s have been seen as the first step toward full brain uploading. But in order to actually scan all that gunk in your head and make it make sense, some scientists believe we’d need the help of a superintelligent AI.
With tech giants in silicon valley throwing billions into AI researchthis future could arrive sooner than you’d expect.
JOIN THE CONVERSATION
Anyone can read Conversations, but to contribute, you should be a registered Torstar account holder. If you do not yet have a Torstar account, you can create one now (it is free)