Skip to content
Home » New Hope for Humans in an AI World: Louis Rosenberg (Transcript)

New Hope for Humans in an AI World: Louis Rosenberg (Transcript)

Louis Rosenberg at TEDxKC

Here is the full transcript of technologist Louis Rosenberg’s TEDx Talk: New Hope for Humans in an AI World @ TEDxKC 2017 conference.

Listen to the MP3 Audio: New hope for humans in an A.I. world by Louis Rosenberg @ TEDxKC

YouTube Video link:

Louis Rosenberg – Technologist, Prolific Inventor, Entrepreneur, Writer

Wow! There’s a lot of smart people in this room. Really smart.

After all, every single one of you has about 85 billion neurons in your heads. And every one of those neurons is connected to thousands of others, which means there is over 100 trillion connections inside every human brain. It’s the system formed by these interconnected neurons that make you who you are. Not just your intelligence, your personality, your creativity, your values and morals, every one of us is a complex system of neurons. It’s remarkable.

After all, we humans are pretty damn smart. We’ve created art, science, literature, The Simpsons. What’s also remarkable is how quickly we got so smart.

If you go back just 2 million years, our brains were one third the size. We didn’t even have language, let alone math and physics. But then the evolutionary pressures got so intense that for us to survive, we had to become the smartest creatures on the planet.

So here we are, the most intelligent species on Earth. Congratulations!

Unfortunately, things are about to change. That’s because an alien intelligence is heading towards us at breakneck speed, and it will challenge our position as the intellectual top dog.

Many experts predict it will get here in 50 years, some say it’ll get here in 20. And let’s just be straight, we have no reason to believe it will be friendly. It’ll have its own values, its own morals, its own self interests. And if it behaves anything like we do, it’ll put its own self-interest first to the detriment of all other creatures it encounters.

No, this alien is not rocketing through space. It will be born right here on earth. A sentient AI that will emerge in a research lab at Google, or Facebook, or a top university.

How do I know it’ll emerge? I know because I have billions of neurons in my head. And I know that research labs are spending billions of dollars creating artificial brains composed of billions of artificial neurons. Mother nature has already proven that this approach works. It will happen. An alien will arrive, and it will be smarter than us.

And when that happens, it will be our next big evolutionary pressure point. One that again requires us to get smarter, much smarter, or lose our position as the top intelligence in our environment. But this time we might not have a few million years. We might only have a few decades.

I know there’re those who say we just need to be careful, and we can prevent this alien from ever hatching in a lab. There’re others who say we just need to put controls in place to be sure that this alien is helpful and friendly, eager to do our bidding. But honestly, we humans have a pretty bad track record for containing dangerous technologies.

So what do we do? Personally, I’ve been worried about this problem for 25 years, ever since I took my very first course on artificial neural networks back in grad school. And for a long time, I didn’t think there was a solution.

A decade ago, I even wrote a science fiction graphic novel called Upgrade that outlines exactly how I see humanity, designing ourselves out of existence by building the first sentient AI.

And how did we let ourselves get so vulnerable to this fictional AI? We gave it control of our infrastructure. We even allowed it to deliver us our food.

A decade ago, people told me that AI delivery drones were far-fetched. Now it seems inevitable. Needless to say, I was sure we were doomed.

But then I asked an innocent question. “How have other species handled evolutionary pressures with a need to get smarter, much smarter fast?” It turns out the answer goes all the way back to the birds and the bees and fish and ants. All of these creatures have evolved methods of amplifying their intelligence by thinking together in systems. It’s why birds flock and fish school and bees swarm. They’re smarter together than alone.

No, I am not talking about crowd sourcing like we humans do, by taking votes and polls and surveys. I’m talking about forming systems. Real-time systems with feedback loops so deeply interconnected that a new intelligence forms. An emergent intelligence with its own personality and intellect.

I’m talking about forming a Hive Mind. Biologists called this Swarm Intelligence, and it’s a natural step in the evolution of most social species.

I like to think about it this way. A brain is a system of neurons so deeply connected that an intelligence forms. A swarm is a system of brains so deeply connected that a super intelligence forms.

Simply put, a swarm is a brain of brains, and it can be smarter than any individual member. Let me give you an example – honeybees. There’s about 10,000 bees, and they have a very important problem to solve. They need to find a new home to move into. That new home could be a hollow log or the hole in a side of a building or if you’re unlucky like I was, a crawlspace in your garage.

Now this sounds simple, but this is a life or death decision that could impact the survival of the colony for generations. To solve this problem, the bees send out hundreds of scout bees that search a 30 square mile area and find dozens of candidate sites. That’s the easy part. The hard part is that they then need to pick the best possible solution from all the options that they’ve discovered.

Now here is the rub. A honeybee has a tiny brain. It is smaller than a grain of sand and has less than a million neurons. You have 85 billion neurons. So however smart you think you are, divide that by 85,000, and that’s a honeybee. You probably wouldn’t want a honeybee picking a new home for you.

And yet honeybees are very discriminating house hunters. They need to find a new home that’s big enough to store the honey they need for the winter, that’s ventilated well enough to stay cool in the summer, that’s insulated well enough to stay warm on cold nights, that’s protected from the rain but also near good sources of clean water, and of course, it needs to be well-located near good sources of pollen.

This is a complex multivariable problem and to maximize survival, the bees need to find the best possible solution across all of these competing constraints, and remarkably, they do it.

Biologists have shown that honeybees pick the best possible solution over 80% of the time. If you were a human CEO, trying to find a perfect location for a new factory, you’d face a similarly complex problem, and it would be very difficult to find the optimal solution, and yet honeybees can do it.

Let’s think about that. A single bee has a brain so tiny it can’t even conceive of the problem, but when they think together as a system, they can solve it so accurately that a human brain would be hard-pressed to match them.

So how do they do this? By forming a swarming intelligence, a brain of brains that combines the knowledge and wisdom and insight and intuition of the group and converges on a single unified decision.

I know what you’re thinking. Really? These are honeybees. How do they express opinions? How do they converge on solutions? Remarkably, they do it by vibrating their bodies. Biologists call this a waggle dance because to us, it looks like the bees are dancing. But really they’re generating signals that represent their support for the various home sites under consideration.

By combining these signals, the bees engage in a multi-directional tug of war, pushing and pulling on the decision until they converge on the one solution that they can best agree upon, and it’s usually the optimal solution. And unlike us humans, the bees don’t entrench, they don’t fall into gridlock, they don’t settle for a bad solution that nobody’s happy with. They find the solution that’s best for the group as a whole.

I point this out because the phrase “Hive Mind” often gets a bad rap, implying a group of mindless drones who can’t think for themselves. That’s certainly how science fiction portrays it. But honestly, it’s just not true.

A hive mind is nature’s way of aggregating the diverse perspectives of a population and maximizing their collective wisdom. And let’s be honest. We humans are pretty smart as individuals, but in groups, we’re not always that wise. That’s because groups don’t do a great job of combining their diverse perspectives.

We use votes and polls and surveys. The problem is polls are polarizing. They drive us to entrench, exposing and reinforcing our differences, but doing nothing to help us to find common grounds.

A swarm is the opposite. It’s flexible and dynamic, revealing where we agree most. In other words, swarming doesn’t just make a species smarter, it makes the species wiser.

Now we shouldn’t feel bad that honeybees are so far ahead of us. They’ve been doing this for 30 million years. We humans have only been around for 200,000. Give us a few million years and we’ll catch up, which brings me back to the problem I started with.

We may not have a few million years. Our next big evolutionary pressure point may hit us within just a few decades, which is why I asked myself a big question. Why can’t we humans amplify our intelligence now? If birds and bees and fish can form a brain of brains, why can’t people do it? That’s what I wanted to know.

So a few years ago, I founded a unique artificial intelligence company called Unanimous AI. We build hive minds. So let me show you what we’ve been up to.

This is a natural swarm. Over the last few years, we’ve been modeling how swarms like this amplify the intelligence of groups, and we’ve been using those models to create the algorithms and interfaces to allow humans to form similar swarms online. This is a human swarm. It’s about a 100 people, all working together to move that glass puck.

Each of the little magnets you see is controlled by a person logged in from somewhere around the world. By moving their magnets, they’re varying their opinions in real time, pushing and pulling on the system in much the same way that honeybees do by waggle dancing.

So how can a swarm like this answer questions? Let me give you some examples.

Last year CBS Interactive challenged us to predict the Kentucky Derby. And not just the winner, the first four horses in order. In horse racing that’s called the superfecta, and last year it went off at 540 to 1 odds. Now, we didn’t know anything about horse racing, we had never done anything like this before, but we were game, so we formed a swarm of 20 horse racing enthusiasts. Not experts, just enthusiasts, and we had them think together as a system to predict the winners of the race.

So here’s a swarm predicting the first place finisher, Nyquist. We did the same thing for second place, third place, fourth place, and then the reporter wrote a story and published the results. She even went to the Kentucky Derby, placed a bet on a superfecta, and tweeted out her ticket, which put some pressure on us.

So how did we do? We nailed it. Anybody who had placed a $20 bet – Thank you. So anybody who had placed a $20 bet on our predictions would have won $11,000. I placed a $20 bet and won $11,000. The reporter placed a bet, a bunch of her readers placed bets. One of her readers reported winning $50,000. Now, this is remarkable.

What’s even more amazing is that of the 20 people in the swarm, not a single one of them got all four horses right on their own. In fact, by taking a vote, they only got one horse right out of four. But by thinking together as a swarm, they were perfect. That’s the power of swarm intelligence.

Let me give you another example. Last year, Newsweek challenged us to predict the Oscars. So, we formed a swarm of 50 movie fans. Not experts, just regular fans, and we had them think together as a system to predict each of the categories.

So here’s the swarm predicting best actress. And you can see them pushing and pulling on the system moderated by our AI algorithms and converging on Brie Larson, and we did the same thing for each of the different categories, and then Newsweek published our predictions, which put some pressure on us.

So how did we do? Let’s compare the individuals to the swarm. So those 50 movie fans as individuals were 40% accurate – that’s not great, but predicting the Oscars is hard.

By taking a vote those 50 movie fans were 47% accurate, which is a little bit better, but by thinking together as a swarm, they jumped all the way up to 76% accurate — that’s almost double the accuracy of the individual members.

And what’s even more amazing is we can compare to the average professional movie critic, which was 64%. In other words, by thinking together as a swarm, these average movie fans became an artificial super expert that was able to beat the New York Times, the LA Times, Rolling Stone, Variety.

Now, these are just two examples, but study after study showed that when people think together in swarms, they can amplify their intelligence by 20%, 40%, 60%. And that’s using current technology. The long-term potential is likely much greater.

After all, if a swarm of bees can solve complex problems that would challenge a human brain, a swarm of humans should be able to solve problems that we can’t even conceive. We should be able to form a true super intelligence.

And because the building blocks are people, tapping not just our knowledge but our values and morals and sensibilities, the resulting super intelligence will not be alien, it will be human, just smarter and wiser. We might be able to solve some of the hardest problems we face, like poverty and inequality and sustainability.

So the next time somebody uses the phrase “Hive Mind” as a negative, remember this: if honeybees could observe how we humans make big group decisions, like electing our leaders, or resolving our conflicts, or planning for our future, they might think we’re the primitive ones — but not for long.

Thanks.

 

Related Posts

Reader Disclosure: Some links on this Site are affiliate links. Which means that, if you choose to make a purchase, we may earn a small commission at no extra cost to you. We greatly appreciate your support.

Multi-Page