A recent blog post from Sam Altman, President of Y Combinator, warns us about the possible consequences of machine intelligence. What is “machine intelligence”, exactly?
“Machine intelligence is a term that [Altman] created to describe the growing automation of computing power,” said Luis Alvarez, CEO of Alvarez Technology Group, on PowerTalk Radio, 101.1. “We think of computers as sophisticated calculators, they sit there and follow instructions that we give them, but more and more the software and hardware advances in the last few years have combined to give computers the ability to essentially come to their own conclusions.”
Machine intelligence has already found a place in pop culture. The Terminator film series popularized the idea of manmade intelligent machines turning against us, wiping out most of humanity with a barrage of nuclear missiles, and more recently we were treated to a more docile, real-life intelligent machine on the game show Jeopardy!:
“If you’ve ever watched the IBM computer play on Jeopardy!, you would have experienced what it is like, the idea that a computer could hear words and then translate those words into something you’d understand and then answer a question that was posed to it based on its consciousness, if you will, or its ability to reason.”
“The Watson computer [used on Jeopardy] was amazing, and since then, and remember that was three years ago, computing power has increased even more. Machine intelligence is this concept that computers will become somewhat independent, able to reason and come to conclusions with human intervention… autonomous to a large degree.”
How close are we to seeing autonomous machines in our everyday lives?
“We already have computers today,`” Alvarez continued, “that can take the code that was used to get them to function and improve it by just going through the coding they were given: maybe correcting mistakes that the programmer put in there, or just making their code more efficient. And that’s used all the time. You know, a human programmer will complete a program and then feed it into another program and say ‘hey, fix my program and make it better.’ At some point, computers will just be doing this on their own. ”
Altman isn’t the only high-profile tech figure concerned about machine intelligence:
“Elon Musk, Stephen Hawking, some very famous folks out there are sounding the alarm that we shouldn’t rush into a world where computers are independently thinking for themselves, because we don’t know what the results are going to be.”
“Without any sort of control, it’s not so much that the machines will be bad in and of themselves. But what’s to keep a bad actor, a terrorist, or some cybercriminal from created a ‘bad’ robot, if you will, or a bad computer that does bad things, because that’s the way it’s programmed and it becomes independent and goes out there and it starts doing its own thing.”
The idea of criminals getting a hold of this technology is disturbing indeed, and the potential legal uses are troubling as well. Police responses to public protests often turn ugly with humans in charge, and if machines began being using in riot control situations, it’s easy to imagine the results being even uglier.
“There is some valid concern that as we approach this era where we are going to have thinking machines, that can reason for themselves and make conclusions and won’t need humans to guide them, we have to tread consciously Imagine a robot that’s making clear, uncompromising decisions without any sort of emotion or consciousness, just using pure logic. It would not be a pretty sight.”
How can we stop these worst-case scenarios of machine intelligence gone wrong from happening?
“What Altman is calling for is regulation at the international level to guard against this sort of thing, to create some sort of regulatory regime, if you will, that will monitor [countries with machine intelligence], just like we monitor countries today that develop nuclear power, that are developing these advanced computing skills to make sure they don’t use them to do harm to society.”
Sometimes Dumb is Better than Smart
Another article, “Decentralization: Why Dumb Networks Are Better”, posted by Andreas Antonopoulos in the Foundation for Economic Education blog, highlights a similar topic, arguing for what he calls “dumb networks”.
“He makes a very good argument,” said Alvarez, “what he’s really talking about is you have certain networks that are ‘smart’, networks that provide all of the computing and thinking that goes into getting results from edge devices, the little things that you access that network with that they don’t do any of that.”
“Conversely, dumb networks, really all they do is just connect devices. The devices themselves are where all the computing and the reasoning and the work take place. Think of working at your office, where your computer’s connected to a server and to the other PCs and printers on your network. What [Antonopoulos] argues is that a dumb network is much more likely to lead to innovation and to foster creativity, as opposed to a smart network where everything is done for you.”
Smart networks are nothing new; think of what the phone system used to look like:
“Remember that back then you couldn’t own a phone, the phone was supplied to you by Ma Bell. When you subscribed to their service they gave you a few options but basically you rented, it’s kind of like cable service today. You know, you don’t own the cable box, the cable box is provided to you by the carrier.”
And indeed, the phone companies did everything they could to stifle innovation, using regulations, litigation, and lobbying to try and slow down the development of the modem, which moved intelligence to the edge devices instead of the edge itself. To help prove Antonopoulos’s point, Alvarez highlighted the best example we have of a dumb network today:
“The ultimate dumb network today: the internet.”
Has there ever been a development that led to more innovation than the internet? While the competition may hurt some businesses, dumb networks clearly lead to more innovation and are better for society as a whole.