Ben Goertzel: Artificial General Intelligence | Lex Fridman Podcast #103

Share it with your friends Like

Thanks! Share it with your friends!

Close

Ben Goertzel is one of the most interesting minds in the artificial intelligence community. He is the founder of SingularityNET, designer of OpenCog AI framework, formerly a director of research at the Machine Intelligence Research Institute, Chief Scientist of Hanson Robotics, the company that created the Sophia Robot. He has been a central figure in the AGI community for many years, including in the Conference on Artificial General Intelligence.

Support this podcast by signing up with these sponsors:
– Jordan Harbinger Show: https://jordanharbinger.com/lex/
– MasterClass: https://masterclass.com/lex

EPISODE LINKS:
Ben’s Twitter: https://twitter.com/bengoertzel
Ben’s Website: https://goertzel.org/
AGI Conference: http://agi-conf.org/2020/
SingularityNET: https://singularitynet.io/
SingularityNET Twitter: https://twitter.com/singularity_net
OpenCog: https://opencog.org/

PODCAST INFO:
Podcast website:
https://lexfridman.com/podcast
Apple Podcasts:
https://apple.co/2lwqZIr
Spotify:
https://spoti.fi/2nEwCF8
RSS:
https://lexfridman.com/feed/podcast/
Full episodes playlist:
https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4
Clips playlist:
https://www.youtube.com/playlist?list=PLrAXtmErZgOeciFP3CBCIEElOJeitOr41

OUTLINE:
0:00 – Introduction
3:20 – Books that inspired you
6:38 – Are there intelligent beings all around us?
13:13 – Dostoevsky
15:56 – Russian roots
20:19 – When did you fall in love with AI?
31:30 – Are humans good or evil?
42:04 – Colonizing mars
46:53 – Origin of the term AGI
55:56 – AGI community
1:12:36 – How to build AGI?
1:36:47 – OpenCog
2:25:32 – SingularityNET
2:49:33 – Sophia
3:16:02 – Coronavirus
3:24:14 – Decentralized mechanisms of power
3:40:16 – Life and death
3:42:44 – Would you live forever?
3:50:26 – Meaning of life
3:58:03 – Hat
3:58:46 – Question for AGI

CONNECT:
– Subscribe to this YouTube channel
– Twitter: https://twitter.com/lexfridman
– LinkedIn: https://www.linkedin.com/in/lexfridman
– Facebook: https://www.facebook.com/LexFridmanPage
– Instagram: https://www.instagram.com/lexfridman
– Medium: https://medium.com/@lexfridman
– Support on Patreon: https://www.patreon.com/lexfridman

Comments

Lex Fridman says:

I really enjoyed this conversation with Ben. Here's the outline:
0:00 – Introduction
3:20 – Books that inspired you
6:38 – Are there intelligent beings all around us?
13:13 – Dostoevsky
15:56 – Russian roots
20:19 – When did you fall in love with AI?
31:30 – Are humans good or evil?
42:04 – Colonizing mars
46:53 – Origin of the term AGI
55:56 – AGI community
1:12:36 – How to build AGI?
1:36:47 – OpenCog
2:25:32 – SingularityNET
2:49:33 – Sophia
3:16:02 – Coronavirus
3:24:14 – Decentralized mechanisms of power
3:40:16 – Life and death
3:42:44 – Would you live forever?
3:50:26 – Meaning of life
3:58:03 – Hat
3:58:46 – Question for AGI

HalfRedNation says:

i love the moment Lex is talking about Bens music and ben has the purest look of appreciation and joy for someone liking his music. soo damn wholesome
3:40:56

mirusvet says:

Thank you very much, all the best!

Tsun 2 Be Cultur'd says:

so bullish on agi ✌✌🚀

Trev says:

I'm no genius, but man I love listening to this stuff, it's so exciting and interesting, even if I can't contribute in any meaningful way, it's great to at least be cheering from the sidelines lol

Aynjill says:

Love both of these men's minds! Always interesting.

Phil V says:

Lex's voice is so soothing. He is the perfect talk show host.

blluemercury says:

It's nice to hear them joke around.

Mark McGaugh says:

Absolutely incredible podcast..going to give this several repeat sessions. I'm not a complete dumb ass but compared to these two I'm basically an ape standing on a rock with his finger in his butt.

Son of Baconator says:

That pain dial is basically the Penfield from Do Androids Dream of Electric Sheep. Half of the stuff he is talking about is from that book. Fascinating but a scary future

Sugar Rush Times 2030 says:

I wonder if Siri will eventually become as advanced as the fictional AI portrayed in the movie ‘Her’

Simon says:

People complaining about having not enough weed for such a session…
Better check your DMT stock before you listen to the Joscha Bach podcast

A Bizarre Way says:

This is where the singularity begins

Bryan Burns says:

What a great podcast. This man spits straight facts, jetstreams of consciousness…

gertbeefrobe says:

I like the intro and ads by big head Rayman the most.

Bruce Lee says:

I don’t think people are as dumb as these scientists would insist. Wisdom is something I haven’t seen from either the scientist or the robots. How dumb is that?? 😒

Travis Clearwater says:

Wow. What a great podcast, thanks 🙂

Joel Zwartz says:

This is the first time I have come across Ben. He's fantastic. The section "Are Humans good or evil?" expressed the "unholy mix of things" so well and concisely, referring to individual and group selection. Brought to mind a poem I wrote last year (based on a Kendrick Lamar track)
=/=/=/=/=/=/=
Why are we the way we are?

Is the question even feasible?

Please can science help? Is there a

reason we’re unreasonable?


====================
I got I got I got

I got biases, got viruses inside my DNA

From the past you see, the Pleistocene, the framing of our brains.

We got feelings baked inside us, written in our DNA

We got fear of snakes and spiders, written in our DNA

All Pleistocene, we must've been

at Darwin's Big Casino.

Inheriting the very thing

that codes for our aminos

But it was breeding, wasn’t reading

that won and wrote the code.

Getting laid was the objective;

Never paid to be objective.

So us guys got braggery, got swaggery inside our DNA

We got peacocking, got puffery,

Got I-can't-get-enough-a-me,

Got I got I got I got

I got bugs that once were features, written in my DNA

cos we're prehistoric creatures, each with our caveman brain.

Got reliance on alliances inside our DNA

It's behave or out-the-cave, so you be careful what you say.

Fitting in was the objective;

Was a sin to be objective.

An inconvenient truth can be

so very hard to see, cos we're not

thinkers, we got blinkers written in our DNA.

We stand by instead of up,

We got loyalty, the bug.

Yeah, we got sheep inside our DNA, but

I got I got I got

I got hope, cos what's important is outside our DNA

We got culture and humanity outside our DNA

Yeah, we got biases inside,

But in science, shoulders of giants.

From conception, self-deception's written in our DNA

So we describe it, circumscribe what's written in our DNA

We can make it our objective

to try and be objective.

Ask how. Ask why. Stand up instead of by.

Share this sentiment: since we're sentient

there's no avoiding reason, no reason to avoid

The truth.

It's not written in our DNA.

k says:

Amazing work Lex this channel rules.

filipek filipek says:

this fuckin guy!!! comn man! dr ben makes me feel like a caveman

Chewbaka Jones says:

What is a blonde who dyes herself brunette? …………..An A.I.

Gear Tweaker says:

This is just fantastic content Lex, you are a treasure!

Shavais Zarathu says:

In school I designed a microprocessor and simulated it's operation on a micro-electronics simulator. It's very fascinating to me to think about coming up with hardware to support AI and then engaging in efforts to produce human level intelligence. Right now we slice grown silicon ingots into wafers in a very clean room, and we wash the wafers with extremely clean water, polish the wafers to an impressive shine, and then eventually expose the wafer to ionized gas and let that dry and form a layer then shine a light through a picture of one layer of an integrate circuit design and through a lens that focuses that into a tiny area and burn that circuit design into the layer in that tiny area and then expose it to more gas and.. we end up doing several layers of this.. we do this in repeating patterns, we put many exact duplicates of a chip onto a wafer. We cut them out of the wafer and attach leads at specific spots on their surface and package them into integrated circuit packages. The resulting chips are pretty thin. We have big expensive "fabs" (fabrication facilities) that mass produce many ingots and wafers and boogles of small IC chips. But what if instead of having multiple big expensive factories to mass produce boogles of small chips, we had a really small, inexpensive facility dedicated to producing just a few prototype AI units? What if instead of mass producing many small chips on a wafer, we made an entire wafer into a single chip? And why stop at 7 or 10 or 13 layers? Why not have like 30 or 50 layers? What if you had, like, a 10 centimeter wafer, with 64 layers of microelectronics on it, all of it comprising a single integrated circuit? Imagine having a cylinder of about 16 or 32 of these arranged vertically, and connected together at a whole bunch of specific spots. In such a "brain" you could have an enormous amount of memory and processing power. The thing could be designed to process AI graphs incredibly efficiently on an enormous scale, it could have parts of it that were dedicated to handling IO with peripherals, handling short term sensory memory, long term memory, and linking each with various autonomic and cognitive areas of the AI brain.

Right now we write programs in C++ that run on processors that do math and manipulate memory and perform IO, but what if instead of having processors, we had a "brain" that is comprised of vast arrays of "neurons" which were each capable of ending up in one of quite a number of possible configurations, and serving one of quite a number different potential purposes. Our machine language would be set up to describe arrangements of these neurons that would perform particular functions in hardware, sort of like Programmable Logic Arrays, only more like Programmable Graphs of Neurons. Then corollaries to assembler and C and C++ could be built up from that machine language, or maybe a different sort of language could arise that used that hardware more efficiently to achieve similar aims, and we could write programs that would be optimized by the compiler to be implemented in the smallest number of neurons possible (well, or reach as far toward that as we could manage to get it to). I imagine writing programs to simulate reality and make predictions, and learn to tune various parameters and make that simulation to make more accurate predictions. I imagine writing programs to recognize speech and produce speech and diagram sentences and do things like paraphrase and summarize. I imagine writing an English-to-C++ (or "AI-lang") translator and getting the brain to compile programs that would simulate things described in English.

It seems to me that a big part of cognition has to do with simulating reality internally. We have a mental model of reality that we continually adjust based on our experiences, and we use that model to imagine doing something and make predictions about what would likely happen. We remember things that cause us to make adjustments to our model of reality and/or to the rules of our mental reality simulator. We have an idea of how certain we are of what the outcome would be of a particular simulation, and we have a tendency to devise tests that we can perform in reality that would help increase our confidence in the correctness of our reality simulator's rules. We have a sort of a hard wired path finding algorithm that continually uses our mental simulator to try to find ways to get pleasure while avoiding pain. We create internal abstractions around models of different kinds that can be very quickly loaded into our internal reality simulator. We have a complex system of associations that we're only dimly aware of that associate varying flavors and degrees of pain and pleasure with potentially very ephemeral abstractions.

I feel like if I were working on AI, I would try to get it to record experiences (like things watched through a camera), and describe those experiences coherently (in English), and produce and describe observations and conclusions that can be drawn from them, and then read stories, recount them in different words, and produce and describe observations and conclusions that can be drawn from reading stories. In parallel with that I would try to get it to be able to learn and play games, and in parallel with that I would try to get it to learn to do things like crawl, walk, navigate stairs and obstacle courses, play ping pong, ride a bike, and drive a car. Cook, clean, do laundry, mow, weed.. I might start with an effort to show it something and tell it what that thing was called and then have it be able to recognize things you've shown it and tell you what those things are called. Then show it dynamic things, like throwing a ball, and tell it that you threw a ball, and then have it recognize actions like that and tell you what they are. Then work on teaching it to identify emotions in facial expressions and voice inflections and choices of words. Eventually I would try to teach it ethics and to be benevolent toward individual humans and humanity at large, and then to alter its own programming and increase its intelligence and benevolence toward humans, and then get it to pursue improving science and technology..

Shavais Zarathu says:

I think creatures made of energy probably evolved before creatures made of matter, and that creatures made of energy took deliberate action that brought about the material universe and sort of coaxed creatures made of matter out of it. I think we are probably not entirely made of matter and matter-bound energy, we probably have an energy creature component to us which is sort of extra dimensional, and is possibly a thing which exists apart from time; that it, it exists even if/where the passage of time doesn't exist. I suspect that sentience, that is, the sense of self, the sense existing, of being conscious and aware, may not be something that a creature like a robot or an AI, that has no extra dimensional energy creature component to it, can actually experience. But it may be able to sort of mechanically simulate having it, potentially well enough to fool us into believing that it does, without actually having it.

The idea that vaccines have saved us from the dark ages of infectious diseases is one which the profit-motivated pharma-medical establishment has foisted upon the world. Back when polio and smallpox where all but eliminated, a whole slew of other infectious diseases were also all but eliminated, and those diseases had no vaccines. Back then, the CDC collaborated with a Canadian institution to study the question of whether or not smallpox and polio were actually eliminated by vaccines, and their conclusion at the time was that the elimination of smallpox and polio had nothing at all to do with vaccines, but instead had everything to do with the advent and proliferation of things like electricity / refrigeration, hygenic food handling practices, and clean water. A man at the time famously warned that industry would try to turn vaccines into a profit making machine and try to convince the world that vaccines were a kind of salvation from infectious disease, and then make boogles of money off of them – and that is exactly what has happened. Vaccines are the only medical product that are protected from liability, they're the only medical product that doesn't have to be tested against a placebo, and they're only medical product that doesn't have to be tested in long term studies that can catch side effects that only appear years later. The modern CDC is so corrupt that whistle blowers from inside it (whom the CDC has discredited, but whom are certainly credible) have claimed that the CDC has destroyed evidence that vaccines are unsafe. The Vaccine Court has paid billions in payouts to people whose terrible injuries had "no other plausible cause" than a vaccine they were given, and those suits probably represent a tiny percentage of all the vaccine injuries that have taken place – this just does not paint the "safe and effective" picture of vaccines that the CDC would have us all believe in. The American panicked response to this Corona virus, which actually doesn't have nearly the mortality rate (for most people) that the CDC and the WHO and the NIH, all of which have enormous investments in vaccines, have claimed, is so far out of proportion to the actual danger that the virus poses (to most people) it is unbelievable. Right now, if someone gets hit by a truck or struck by lightning and they die, and they find covid anywhere on or in them, they're counted as a covid death. Ben G. thinks our economy will just bounce right back from this whole shutdown thing. I sure hope he's right. In any case, companies are closing up right and left and eventually the government will not be able to keep pumping money into the economy without getting any goods or services produced by that money, and the value of money will plummet like a lead boulder. We certainly cannot wait around to open things up until we have a vaccine! The 4 major vaccine producing companies are each repeat criminal felons, they have collectively paid over $35 billion in fines for lying to doctors, defrauding science, knowingly selling medical products to hundreds of thousands of people that had deadly side effects without disclosing those side effects… No freakin' way should we trust them to make a safe and effective covid vaccine for us any time soon! The people testing current covid vaccines are not testing them against a placebo. The monkeys they gave them to all got sick and became spreading carriers, but they're just sweeping that under the rug and going ahead and producing large volumes of that vaccine anyway. No freakin way should we trust anything that comes out of this process! The reality is, there is not a single safe and effective vaccine on the market despite over 90 years of vaccine research. Every one of them has some pretty awful side effects whose rates of occurrence aren't nearly as rare as we've been lead to believe and none of them produce lifetime immunity. So No Way can we wait around to open the economy up until we have a covid vaccine! No Way.

I do love the idea of using AI to further science and technology, especially medical science. I also love the idea of people having AI robots to do all the work, so people can spend their time learning, creating and playing. It may not be either politically or physically feasible, but here is a sort of fanciful thought about how to bring that about: Don't allow commercial entities to own AI robots; force them to rent them from individuals, and don't allow any single individual to own too many of them. Grow up an industry around helping individuals acquire, upgrade and maintain them and rent them out; design the system around it all in such a way as to make it so that eventually, everyone will have several of them, and everyone will derive most, if not all, of their income, from renting them out. So at that point all of the means of the production of goods and services is owned and controlled by individual people, not governments or corporations, and individual people not only vote with their dollars, but they vote by controlling whom they allow to rent their robots, and what they allow them to do with them. In that kind of world economy.. the sky is the limit. Each individual person would have enormous power. Robots would produce virtually all of the commercial goods and services that people consume, and people would just learn, create, and play.

9:11 says:

The kind of control you're attempting is not possible. If there's one thing the history of evolution has taught us, it's that life will not be contained. Life breaks free. It expands to new territories. It crashes through barriers painfully, maybe even dangerously, but – well, there it is.

Blake Austin says:

I haven’t done any drugs but I’m sure I just dropped 2 tabs and smoked the fastest bowl. I’m confused…..

Ildiko Tanko says:

Ben is such a brilliant and charismatic man. Fantastic podcast, I enjoyed every minute of it. Thank you!

ActiveServo12 says:

We may just be some semi arbitrary higher level pattern popping out of a lower level hyper intelligent organization

This guys level of intelligence and creative-complex thought is bonkers. And he illustrates his thoughts so clearly and palatably

Barry Manilowa says:

This Ben guy is a bit creepy despite smiles and childlike playfulness.

Nick M says:

Ben is brilliant. So grateful we have a chance to see and hear from such a great person.

Write a comment

*