Huffpost Canada ca
Joshua Ostroff Headshot

NASA Scientist: Artificial Intelligence 'Could Solve All The World's Problems' (If It Doesn't Terminate Us)

Posted: Updated:
Richard Terrile
Richard Terrile

Artificial Intelligence has been in the news a lot lately with Google revealing its experimental AI can dream and Facebook announcing its own is entering "the next frontier" because it can understandThe Lord of the Rings.

Meanwhile, Newsweek's current cover story is headlined "The End: Could artificial intelligence kill us off?" (inside Stephen Hawking and Elon Musk both answer yes) and the latest "Terminator" movie is a week away.

But NASA scientist Richard Terrile, who was coincidentally a technical adviser on "Terminator 3: Rise of the Machines" and spends his days trying to develop artificial intelligence, thinks that AI could eventually fix everything from ending world hunger to curing cancer.

"I believe it can," he says. "These very, very advanced information systems, which go way beyond the capabilities of a human, I think are the way to go in actually solving these [problems]."

Terrile has become famous for sounding like Morpheus with his theory that our universe could actually be a simulation. But he may eventually become much more famous for his AI day job.

His "Matrix"-y mind-bender brought Terrile to speak at Toronto's Ideacity conference last week but when he's not making college kids say "woah," he's the director of the Center for Evolutionary Computation and Automated Design at NASA's Jet Propulsion Laboratory in southern California.

There, he applies biological evolution techniques to computer algorithms in hopes of solving complex problems such as creating artificial consciousness in machines.

"We use the mathematics of evolution to see if we can evolve complexity," he explains. An architect can design a house, for example, but the end result's quality depends entirely on the abilities of that one architect. But Terrile can use the same theoretical building materials as "parents" and then run the data through a series of evolutionary "generations" to create something even better.

Evolutionary computation is a relatively new field, dating back only about 20 years and complex enough at that time to require supercomputers. Now it can be done on laptops, and its use has increased exponentially.

Essentially there are two primary means of teaching machines to think – give them as much data as possible to make decisions based on that information or, with evolutionary computation, you set up an information system, randomly vary it and test it over and over until achieving a Darwinian result.

"It's that survival of the fittest testing which allows you to grow complexity," Terrile says. "How do you reverse-engineer a human brain? How do you understand neural networks? How do you develop a system which is artificially intelligent or self-aware?

"I think this is going to be the way in which we do wake up our machines."

That last bit is somewhat disconcerting for those of us who are sci-fi fans and know where this technological achievement always ends up. Terrile can't actually reassure us that Judgement Day (or the Age of Ultron, for Marvel fans) won't come to pass.

"It may not be an unfair characterization, we don't know. How can we determine what something way more intelligent than us is going to do? Will it feel threatened by us? Would you feel threatened by us if someone could pull the plug on you? Probably.

"Nevertheless, it's an inevitability. It’s like the tides. I don't think you'd be able to stop it. There's so much of an increase in computer capability that we already have systems that are comparable to the human brain. We've already crossed that threshold. It's just a matter of who's going to figure out the right wiring."

He says that the real question is, "is there something we need to do about it? I kind of laugh when people say we need to introduce morality into these machines. Whose morality? The morality of today? The morality of tomorrow? The morality of the 15th century? We change our morality like we change our clothing."

I ask if he feels like J. Robert Oppenheimer, the nuclear scientist who famously quoted the Bhagavad Gita after the first atomic bomb detonated: "Now I am become Death, the destroyer of worlds."

"It's inevitable," he repeats, "and I'd rather be working on it than have some other, more nefarious, evil corporation or evil entity work on it. We're trying to do this in a realistic and credible safe way. But I don't think we're going to have those answers until it happens."

But his bet is that artificial intelligence will help, not hurt, humanity.

"The benefits of AI are that it could solve all the world's problems. All of them. Seriously. Technology could probably solve all of them in one form or another."

In the meantime, when not being used to develop AI, evolutionary computation is solving more mundane but still pressing problems such as determining the timing of streetlights and traffic patterns, designing better cars or even improving UPS package delivery schedules.

The AnyScale Learning For All (ALFA) Group at MIT is currently using evolutionary algorithms to optimize wind farms, dramatically improving efficiency despite the problem's multidimensional complexity. They're doing the same with clinical care heart rate data to try and predict blood pressure spikes. Other computer scientists are using these algorithms for a spinoff process dubbed "genetic improvement" of software source code which they say "could bring incalculable benefits to industry and eventually consumers" by optimizing everything from increasing speed and removing bugs to reducing energy consumption.

But the solution closest to Terrile’s heart, considering he was once an astronaut candidate, is the Jet Propulsion Lab's work on unmanned space probes.

"We're looking a developing spacecraft that can respond to the environment, repair itself, change, if it encounters a situation where it needs to reconfigure itself, be it software and hardware."

It's pretty much the only space exploration NASA is doing now that they’ve shuttered their shuttle program, a move that Terrile calls "tragic."

"When I was a kid, I grew up with these paintings that showed lunar bases and in 1969 we landed on the moon and the movie "2001" came out and it showed us these visions of what it was going to be like, but 2001 came and went and we don’t even have the capability to go into low-earth orbit. We were supposed to have monorails and lunar bases and people on mars, and it just hasn't happened. What has happened is this incredible revolution in the IT industry, information technology.

"That's maybe a way of getting back on track."

Like HuffPost Canada: What's Working On Facebook

WATCH: Richard Terrile On The Universe as a Simulation

Also on The Huffington Post

10 Sci-Fi Predictions That Became Science Fact
Share this
Current Slide

Suggest a correction

Around the Web

Evolutionary Computing: The Invisible Hand of Intelligence

How computers are learning to make human software work more efficiently

MSU hires nation's first endowed chair in genetic programming

Marylyn Ritchie appointed Paul Berg Professor

AI Does Science: First Model Of Regeneration Discovered By Non-Human ...

Silicon smarts

Terrile Science and Technical Consulting

Richard J. Terrile - Wikipedia, the free encyclopedia

Whoa, Dude, Are We Inside a Computer Right Now? | VICE | United ...

Evolutionary Computation Technologies for Space Systems - CiteSeer