Voices in AI – Episode 61: A Dialog with Dr. Louis Rosenberg

As we speak's main minds discuss AI with host Byron Reese

About this Episode

Episode 61 of Voices in AI options host Byron Reese and Dr. Louis Rosenberg speaking about AI and swarm intelligence. Dr. Rosenberg is the CEO of Unanimous AI. He additionally holds a B.S., M.S., and a PhD in Engineering from Stanford.

Go to www.VoicesinAI.com to hearken to this one-hour podcast or learn the complete transcript.

Transcript Excerpt

Byron Reese: That is Voices in AI, dropped at you by GigaOm. I’m Byron Reese and at present I’m excited that our visitor is Louis Rosenberg. He's the CEO at Unanimous A.I. He holds a B.S. in Engineering, an M.S. in Engineering and a PhD in Engineering all from Stanford. Welcome to the present, Louis.

Dr. Louis Rosenberg: Yeah, thanks for having me.

So inform me a little bit bit about why do you've gotten an organization? Why are you CEO of an organization known as Unanimous A.I.? What's the unanimous side of it?

Certain. So, what we do at Unanimous A.I. is we use synthetic intelligence to amplify the intelligence of teams somewhat than utilizing A.I. to exchange folks. And so as a substitute of changing human intelligence, we're amplifying human intelligence by connecting folks collectively utilizing A.I. algorithms. So in laymen’s phrases, you'll say we construct hive minds. In scientific phrases, we might say we construct synthetic swarm intelligence by connecting folks collectively into methods.

What's swarm intelligence?

So swarm intelligence is a organic phenomenon that individuals have been finding out, or biologists have been finding out, for the reason that 1950s. And it's mainly the rationale why birds flock and fish faculty and bees swarm—they're smarter collectively than they'd be on their very own. And the way in which they develop into smarter collectively shouldn't be the way in which folks do it. They don’t take calls, they don’t conduct surveys, there’s no SurveyMonkey in nature. The best way that teams of organisms get smarter collectively is by forming methods, real-time methods with suggestions loops in order that they'll primarily suppose collectively as an emergent intelligence that's smarter as a uniform system than the person members could be on their very own. And so the way in which I like to think about a man-made swarm intelligence or a hive thoughts is as a mind of brains. And that’s primarily what we give attention to at Unanimous A.I., is determining how to do this amongst folks, regardless that nature has found out how to do this amongst birds and bees and fish, and have demonstrated over hundreds of thousands of years and a whole bunch of hundreds of thousands of years, how highly effective it may be.

So earlier than we discuss synthetic swarm intelligence, let’s simply spend a little bit time actually attempting to grasp what it's that the animals are doing. So the thesis is, your common ant isn’t very good and even the neatest ant isn’t very good and but collectively they exhibit habits that’s fairly clever. They'll do all types of issues and forage and do that and that, and construct a house and defend themselves from a flood and all of that. So how does that occur?

Yeah, so it’s a tremendous course of and its price taking one little step again and simply asking ourselves, how can we outline the time period intelligence? After which we will discuss how we will construct a swarm intelligence. And so, in my thoughts, the phrase intelligence may very well be outlined as a system that takes in noisy enter in regards to the world and it processes that enter and it makes use of it to make choices, to have opinions, to resolve issues and, ideally, it does it creatively and by studying over time. And so if that’s intelligence, then there’s a lot of methods we will take into consideration constructing a man-made intelligence, which I might say is mainly making a system that entails expertise that does some or all of those methods, takes in noisy enter and makes use of it to make choices, have opinions, remedy issues, and does it creatively and studying over time.

Now, in nature, there’s actually been two paths by which nature has found out tips on how to do this stuff, tips on how to create intelligence. One path is the trail we’re very, very accustomed to, which is by increase methods of neurons. And so, over a whole bunch of hundreds of thousands and billions of years, nature found out that if you happen to construct these methods of neurons, which we name brains, you possibly can absorb details about the world and you need to use it to make choices and have opinions and remedy issues and do it creatively and study over time. However what nature has additionally proven is that in lots of organisms—notably social organisms—as soon as they’ve constructed that mind and so they have a person organism that may do that on their very own, many social organisms then evolve the power to attach the brains collectively into methods. So if a mind is a community of neurons the place intelligence emerges, a swarm in nature is a community of brains which are related deeply sufficient {that a} superintelligence emerges. And by superintelligence, we imply that the mind of brains is smarter collectively than these particular person brains could be on their very own. And as you described, it occurs in ants, it occurs in bees, it occurs in birds, and fish.

And let me discuss bees as a result of that occurs to be the kind of swarm intelligence that’s been studied the longest in nature. And so, if you concentrate on the evolution of bees, they first developed their particular person brains, which allowed them to course of data, however in some unspecified time in the future their brains couldn't get any bigger, presumably as a result of they fly, and so bees fly round, their brains are very tiny to have the ability to enable them to do this. Actually, a honeybee has a mind that has lower than 1,000,000 neurons in it, and it’s smaller than a grain of sand. And I do know 1,000,000 neurons seems like lots, however a human has 85 billion neurons. So nonetheless good you might be, divide that by 85,000 and that’s a honeybee. So a single honeybee, very, quite simple organism and but they've very troublesome issues that they should remedy, similar to people have troublesome issues.

And so the kind of drawback that's truly studied probably the most in honeybees is choosing a brand new house to maneuver into. And by new house, I imply, you've gotten a colony of 10,000 bees and yearly they should discover a new house as a result of they’ve outgrown their earlier house and that house may very well be a gap in a hole log, it may very well be a gap along side a constructing, it may very well be a gap—if you happen to’re unfortunate—in your storage, which occurred to me. And so a swarm of bees goes to wish to discover a new house to maneuver into. And, once more, it seems like a fairly easy resolution, however truly it’s a life-or-death resolution for honeybees. And so for the evolution of bees, the higher resolution that they'll make when choosing a brand new house, the higher the survival of their species. And so, to resolve this drawback, what colonies of honeybees do is that they kind a hive thoughts or a swarm intelligence and step one is that they should acquire details about their world. And they also ship out a whole bunch of scout bees out into the world to go looking 30 sq. miles to search out potential websites, candidate websites that they'll transfer into. In order that’s information assortment. And they also’re on the market sending a whole bunch of bees out into the world looking for totally different potential houses, then they bring about that data again to the colony and now they've the troublesome a part of it: they should decide, they should choose the very best website of dozens of doable websites that they've found. Now, once more, this sounds easy however honeybees are very discriminating house-hunters. They should discover a new house that satisfies a complete bunch of competing constraints. That new house needs to be giant sufficient to retailer the honey they want for the winter. It must be ventilated nicely sufficient to allow them to hold it cool in the summertime. It must be insulated nicely sufficient so it may possibly keep heat in chilly nights. It must be protected against the rain, but additionally close to good sources of water. And in addition, after all, it must be well-located, close to good sources of pollen.

And so it’s a posh multi-variable drawback. It is a drawback {that a} single honeybee with a mind smaller than a grain of sand couldn't presumably remedy. Actually, a human that was that information would discover it very troublesome to make use of a human mind to search out the very best resolution to this multi-variable optimization drawback. Or a human that's confronted with an identical human problem, like discovering the proper location for a brand new manufacturing facility or the proper options of a brand new product or the proper location to place a brand new retailer, could be very troublesome to discover a good resolution. And but, rigorous research by biologists have proven that honeybees choose the very best resolution from all of the out there choices about 80% of the time. And after they don’t choose the very best resolution, they choose the subsequent very best resolution. And so it’s exceptional. By working collectively as a swarm intelligence, they're enabling themselves to decide that's optimized in a approach {that a} human mind, which is 85,000 instances extra highly effective, would battle to do.

And so how do they do that? Nicely, they kind a real-time system the place they'll course of the information collectively and converge collectively on the optimum resolution. Now, they’re honeybees, so how do they course of the information? Nicely, nature got here up with a tremendous approach. They do it by vibrating their our bodies. And so biologists name this a “waggle dance” as a result of to people, when folks first beginning trying into hives, they noticed these bees doing one thing that appeared like they had been dancing as a result of they had been vibrating their our bodies. It appeared like they had been dancing however actually they had been producing these vibrations, these alerts that characterize their help for his or her numerous house websites that had been into consideration. By having a whole bunch and a whole bunch of bees vibrating their our bodies on the identical time, they’re mainly participating on this multi-directional tug of struggle. They’re pushing and pulling on a choice, exploring all of the totally different choices till they converge collectively in actual time on the one resolution that they'll greatest agree upon and it’s virtually at all times the optimum resolution. And when it’s not the optimum resolution, it’s the subsequent greatest resolution. So mainly they’re forming this real-time system, this mind of brains that may converge collectively on an optimum resolution and might remedy issues that they couldn’t do on their very own. And in order that’s probably the most well-known instance of what a swarm intelligence is and we see it in honeybees, however we additionally see the identical course of taking place in flocks of birds, in colleges of fish, which permit them to be smarter collectively than alone.

Hearken to this one-hour episode or learn the complete transcript at www.VoicesinAI.com

 

Byron explores points round synthetic intelligence and acutely aware computer systems in his new e book The Fourth Age: Good Robots, Acutely aware Computer systems, and the Way forward for Humanity.

!function(f,b,e,v,n,t,s)if(f.fbq)return;n=f.fbq=function()n.callMethod?
n.callMethod.apply(n,arguments):n.queue.push(arguments);if(!f._fbq)f._fbq=n;
n.push=n;n.loaded=!0;n.version='2.0';n.queue=;t=b.createElement(e);t.async=!0;
t.src=v;s=b.getElementsByTagName(e);s.parentNode.insertBefore(t,s)(window,
document,'script','https://connect.facebook.net/en_US/fbevents.js');

fbq('init', '1157339287680303');
fbq('init', '172929499778809');
fbq('track', "PageView");


Source link

Read the full article