July 16, 2021

Interesting ideas from the Lex Fridman Podcast

About two months ago I discovered the wonderful Lex Fridman podcast. It has kept me company most of my days since then and has made me pause and think about what was said by Lex or his guests and life in general really about a hundred times, thinking about some of the ideas mentioned long after the podcast was over.

As a tribute to this wonderful intellectual work, since I wrote some of the ideas I found really interesting into my little notebook, I wanted to post them as an article.

I intentionally didn't attribute individual thoughts or paragraphs to their authors (I did attribute them as a group at the end of the article) and there is a reason for that. I've been experimenting with a new concept (at least for me). I noticed that when I read an article by someone and don't know who wrote it or when I listen to a podcast and do not know anything about the person speaking, it's a completely different intellectual experience for me than when I know at least something about that person. Even seeing the picture sets up a certain preconception in my mind. When I know nothing about the person, I am much more open to that person's ideas, listen/read with much less prejudice and just have much less preconceptual junk in my mind influencing my listening in general.

When I listen to Lex's podcast (or any podcast) now, I skip the guest introduction, and let the 'ideas sell themselves'. I might read up on the guest afterwards, but not before the interview and I am trying to do this with articles I read as well (where possible). I have to say it really is a whole new intellectual experience:)

So as a second 'theme' of this article (the first being a simple sharing of the ideas in Lex's podcast I found interesting enough to write in my notebook a day or two after I had heard the podcast), I decided to not say who said what :) 

I did list the names of those who shared these ideas at the end of the posts, because 
1) it is a fair thing to do 
2) you can look them up if you would like to learn more.


When looking for life, if information is a real physical thing, we should be looking at what would its signatures be in matter and how do we recognize those. There are objects in the universe that obviously didn’t have their design encoded in the laws of the universe and we should look for what made for their existence possible.  

Life is a system that patterns structure into matter, not a molecule of oxygen on an exoplanet. 
(Pointing to the fact that we should have a better framework when trying to look for evidence of extraterrestrial life than looking for the presence of molecules associated with life on Earth - Vlad)

Life is maybe much less obvious than we think. It could be in many more forms than we expect. Maybe information from an alien cannot even be copied to us because of how different that being would be from us. 

When I see a person, I don’t see that person as a 4 billion year lineage, yet that’s what that person is and maybe AI will be able to see that feature of us, and if it could see that, imagine what else it could see. 

(In this context, killing a fly or even taking a shower takes on a completely different dimension. Imagine where all the atoms in the water drops in your shower have been in the past 4 billion years and what it took for them to all appear at that one place now. Same goes for the fly, plus all the genetical information that is in it. The fly is the last link in an unbroken evolutionary chain ~4 billion years long. - Vlad)

If extraterrestrial intelligent beings were to visit Earth, would they even notice humans among the millions of other life forms on our planet at first? There are so many organisms here, why should they deal with us first? If they want size, they could go for whales, if they want quantity, they could go for bacteria. Maybe they will think that cities are organisms which come to life during the night (lights come on), they sleep during the day (lights go off). 

Assembly theory is a much better framework when we look for signs of life on distant exoplanets, because it is not contingent to looking for molecules of life on Earth on other planets.
(Assembly theory is a framework used when trying to detect evidence of extraterrestrial life. It measures the complexity of molecules. In other words, when looking for evidence of extraterrestrial life, with Assembly theory, we are not looking for molecules like water or oxygen on distant exoplantes which are related to life on Earth but measuring 'how many steps it takes' to create a given molecule we see. After a certain threshold, certain molecules have to be created with enough steps that their existence cannot be explained without the presence of life  - Vlad)

If it’s hard for life to evolve it will be far from us by definition.

Life it seems, as we see it on Earth, tends to need liquids to evolve and exist. There are not that many liquids out there, that’s why we look at water on other planets. 

If there is life on Venus, we have to be very careful and we should not go there at all. Venus has suffered a lot during its lifetime. If there is life, it is the last survivor of something that might have been a prosperous biosphere and I don't want our first contact with extraterrestrial life to be a massacre. 

I don't know how frequent life is in our galaxy and the universe. If we find life in our Solar system I will adjust my prediction. If life is found in our solar system, I think life exists everywhere where there are conditions for life to exist. But not intelligent, conscious life capable of communication. There were 5 billion different life forms on our planet and we, as the only ones, have reached the level of intelligence we have and that too only for a very small fraction of time and it is possible that we wipe ourselves out.. so.. the odds for intelligent life (intelligent at least as we are) to evolve are 5 billion to 1. We can estimate this only based on one single instance of life - us. It's not much, but that's all we can do. It is also very probable that intelligent life exists only for a very short period of time, because it wipes itself out. So even if life evolves somewhere, for intelligent life to evolve and exist at this very moment simultaneously with us the odds are very small.

Intelligent extraterrestrial life is very rare. There are maybe 3-4 other intelligent life forms in our Milky way. There were maybe 5 billion life forms on our planet and only one - us, managed to reach intelligence on the level we are right now. 
(For reference, there are 100-400 billion stars in the Milky way and each star has one planet on average. -Vlad)

Traveling even to our nearest stellar neighbors is something that lacks meaning for human beings at this point. Going to Sirius with the greatest speed we can achieve right now would take us ~250 000 years, which is ~600 000 generations.  What could be done though is to send robots, send our DNA and let us evolve over there. 


What is driving me is to study how our minds interact with the physical reality around us.

From a physics standpoint I will never cease existing, because time being the 4th dimension, there is forever going to be an (x, y, z, Time) coordinate where I exist, I just can’t access it. We perceive time as flowing because of delta change but time doesn’t really flow. 

I have a colleague who has a theory that consciousness could be contagious - that it exists only between people or people and dogs etc. 

You can’t know whether cellular automata are conscious or not because you are not a cellular automata.

The human reward function is survival and reproduction.
(A 'reward function' is a system of rewards given to a neural network when it is learning something. If it achieves the goal the people who programmed it wanted it to achieve (for example get to the end of a maze), they 'reward' it mathematically so that the neural network knows it achieved the goal and that this is what it should optimize for. - Vlad)

To be pregnant with possibility.

If stated calmly and with reason you are much more convincing. 

What is the meaning of life?

Guest A)

I used to think it was to multiply and preserve life, but now I think that it is enough that an asteroid wipes us out and this meaning is thus rendered useless. Therefore now I think that the meaning of life is to expand the knowledge we have as humans and prepare an 'escape capsule' where all of it could be stored so that in case an asteroid does wipe us out and 100 000 000 million years later another intelligent life form evolves on our planet, they don't have to learn everything from scratch. 

Guest B)

The meaning of life for me is the equation which has the following parameters: how many people.hours will benefit from how I contribute to humanity during my life after I'm gone. 

Guest C)

You first have to zoom out and ask, what is the meaning/objective of the universe. The universe, based on what we know right now, started with the Big Bang and has been expanding since then and based on laws of thermodynamics is lowering its entropy until the entropy is zero. According to some scientists, the objective of the universe is thus to lower its entropy to zero. Evolution is locally making the lowering of entropy more effective (as is the case on Earth for example). We people, as one micro link in the evolutionary chain are lowering entropy locally even quicker and our objective could be to do it as effectively as possible. You could thus argue that, given the above, the meaning of life is to lower entropy locally as quickly and effectively as possible, which can be achieved by a) preserving life (fight for survival) and b) spread life as much as possible (reproduce). (Everything checks out:) - Vlad)

Artificial intelligence:

...it’s an interesting philosophical question, because if we manage to create Artificial General Intelligence and it is only able to ‘fake’ consciousness then we don’t have to worry about its rights but if it's really conscious, then it's a different story.  
(AGI - intelligence that is as intelligent or better than humans in all tasks, not just chess for example - Vlad)

We have no clue how the brain operates. We are 500 years away from understanding it. We are like the Greeks looking at the moon and thinking it would be fun to go there. Let it be clear - we are nowhere near. Musk, god bless him, leads people astray because he talks about things he doesn’t know a lot about. Natural Language Processing, real language understanding, not faking it, is not going to happen in our lifetime.

Machine learning (ML) is the term that should be used not AI. We have no idea what is intelligence. We are not building that. 

ML are systems that learn and make decisions. AI is not happening right now. We are taking data, we are training a machine to do good decisions on that. Economically viably etc. It will look more intelligent but..it’s not intelligence. 

What would be the first question you would ask AGI?

Guest A) 
What is the meaning of life?

Guest B)
What is outside the simulation?
(Referring to the theory that we live in a computer simulation. Vlad)


Marcus Hutter, Michael I. Jordan, Alex Filippenko, Elon Musk, Po-Shen Loh, David Silver, Sara Walker, Clara Sousa-Silva, Katherine de Kleer, Sara Seager

Staying faithful to my new experiment, I didn't want to mention this at the beginning of this article, but: all of the above persons dedicated their lives to the craft they love and are arguably the best or among the best in what they do in the world.

The picture is a random picture I took at Berlin Templehof. I think it adds to the 'no info reveal' theme:)

No comments:

Post a Comment