Yesterday I caught a discussion on the radio about this story, then searched out the article. It's about Elon Musk and Demis Hassabis debating who's project is the most important project in the world. It reads like a script to a sci-fi movie, reality.

[Link: vanityfair.com]

Not sure if there will be pilots for these vehicles from New York. The teachers might not be able to read past the eight grade level. New York threw out the literacy tests for teachers.

True AI would be the biggest breakthrough in the history of mankind, and would likely lead to advances in interplanetary travel (and beyond) that wouldn't be possible otherwise.

I heard that on the radio a few days ago. They had too many teachers failing the test to get a license on the first try so they dumbed down the test. Heck, why not just throw out the test all together?

Maybe, maybe not. The fact that it's artificial doesn't necessarily imply it's unbounded or even superior to our own. It will be different to ours, which may have some of its own advantages and some of its own disadvantages.

Hopefully we could leverage the best of both to our advantage. But don't expect that real AI is going to figure out the answers to nature's riddles that we can't. I doubt it would be remotely as good at that sort of thing as we are, because all of the great discoveries in that area have tended to be bank shots.

And keep in mind also that genetics are likely going to reach the point within that same time frame where we will start having the opportunities to ratchet our own capabilities up significantly as well. That's also sort of scary, but it means that the AI will most likely have more competition that most folks think.

* Sadly, one of the things AI would more likely be very good at is coordinating massive amounts of data to look for patterns and predict things (and who do you think most of that data is going be collected about?)


In my opinion, the big breakthroughs will be neither of those, though one is related to the former. One will be nuclear fusion, assuming we ever get it to work. The massive jump in energy availability with vastly reduced environmental concern would fundamentally change our situation. So many of our limits and costs are current set by the cost and availability of energy.

The other will not be interplanetary travel, but when we can start taking advantage, on a commercial scale, of off-planet resources. As with the energy issue, the huge jump in the ability of now costly resources will have a huge influence on our lives. And again, without environmental concerns. No one cares if an asteroid gets mined out.

And of course that will also go hand in hand with the requirement to do pretty significant engineering in space, which itself will give us the opportunity to take advantage of solar power in a way that we can't on earth, which would get us another big whack of energy availability.

Compared to that, interplanetary travel will be fairly fluffy stuff, though very cool.

OTOH, we may all kill each long before any of this comes to pass.

AI is already here.


Just look at the person behind the fast food counter....

I find it interesting that no one picked up on Musk's opinion that his plan to colonize mars is the most important project ever, because of the threat of artificial intelligence

What part of AI involves people behind a fast food counter? Certainly not intelligence.

Someone I know posted their experience at a McDonald's and a wonderful woman whose name is Cheetoria.

They almost seem real don't they?

about as real as Abraham Lincoln at Disney World's Hall of Presidents

Until they speak and remove all doubt.

Reminds me of a comment Leonard Hofstadter made on Big Bang when he and Sheldon were arguing about something- "Sometimes, you say things that make me forget that you're a real boy".

I took his post differently and, at least the way I understand it, I would agree with him.

It is not a matter of better it is a matter of simpler. We sent Pioneer 5 (and other space probes since then) out into space in 1960, we sent Pathfinder (and other space rovers since then) in 95 so that sojourner can roll on the planet. You are very limited in what can be accomplished with pre-programmed itineraries and remote control. If we can send probes, rovers and robots with AI that are a lot more independent we can accomplish a lot more. There is no way people can be sent to Trapist-1 to try and find out more about those 7 planets (or any other exoplanet), but you could send something mechanical and if it takes a few centuries to get there, that is not that big a deal. You also can't pre-program it (too many unknowns) or remotely control it (too slow to get a signal back). I used that example because you would need better AI to do it, but even for something like Mars it will help there is a time delay between 4 (best case when Mars and Earth are at their closes) and 25 minutes (worst case farthest and indirect signal) for a signal one way, add some transmission time... things can be much fatser if the rover can make more decisions. it is also why they are built with rudimentary avoidance programming.




agree, both are just toys for rich people.


that could be a big one but like you said that is "assuming we ever get it to work"
and assuming it works the way we would like it to work.....
but the other two are "evolutionary" (i.e. AI is used in the rover for avoidance already, as AI improves the next rover can do a bit more) techs while what you describe is more of an all or nothing (i.e. either it works or it does not)


But wouldn't that be AI and interplanetary travel related or do you mean sending some idiots to Mars is not a big deal, getting them to mine it and send back the stuff is when it becomes a big deal?

Don't know why you would be more excited by that. First there are many movies and books that use the premise of AI destroying the world. Second there are many great minds that believe that AI can be a threat. Lastly if AI is a threat in any way or form it will be a threat on what it controls.

Sure, for exploration, smarter probes will help. I may have misread what he said, thinking he was talking more about the underlying technical and theoretical advances that will be required to make that happen. I don't think AIs are going to top humans in the fundamental sciences department, because all the big discoveries have been very indirect, very intuitive. AIs of course would make excellent lab assistants as part of the process.

And I don't think that the human factor will ever really be removed from space travel and/or off planet mining and manufacturer. You can make AIs as smart as you want, but humans are almost always going to be more likely to trust other humans to make critical decisions. And I think it will be a some time yet before AIs are as smart as people think of them in the sci-fi movie sense as well.

But you don't really need it to be. Devices that are extensions to human capabilities are, to me, more interesting and powerful, than trying to create completely autonomous devices. Computers have their strengths and humans have theirs. A combination of the two, with the human in ultimate control, I think is the way to go. Where it's at all possible at least. In some extreme circumstances, humans would be in too much danger, and robots would be the only way.

BTW, I'm not sure anyone will ever bother sending probes on multi-hundred year missions. So much will have changed before the probe can get there and any information back, that it's probably not going to be seen as being worth it. Most likely, when the probe gets there, there will already have long since been people there via much faster means that became possible a hundred years after the probe was launched. That or the human race will have collapsed and there'll be no one around to hear the reply.

agree with most of what you say.

My point, to put it simply, is that a self driving ATV that stops at the edge of a cliff is all about convenience (i.e. if I am paying attention I can stop the ATV in time), on Mars a self driving rover that stops at the edge of a cliff is needed because if someone from Earth needs to stop it in time the vehicle for chunks of the year will need to barely be moving (i.e. if it takes 1h to give it a stop command because it is getting near the edge of the cliff it can only go inches per hour (be it due to roll and stop or just roll real slow). The farther from earth that you want to explore (and if we are to find an Earth like planet that has life we will need to go real far) the more you need the vehicle to be independent and self reliant.


Don't know if we can or we will, but voyager 1 & 2 where launched in 1977 and roughly 40 years later they are still in commission gathering data as they are at the edge of our solar system (just under one light day away), yes their technology is old and outdated but it takes time to get where they are and so they are extremely valuable, and I am assuming they will continue being used as long as they continue to function.

The nearest solar system to ours is Alpha Centauri and it is roughly 4 light years away and at this point in time it might have interesting exoplanets. If we want to send something there not only do we need to have something that can travel much faster but it would need to run extremely autonomously on a daily basis because any signal would just take to long to be useful in driving it. Imagine we can build a ship that can travel at 0.1 light year we could hen send a probe to get a better look or maybe a ship with some rovers. the issue is that we don't know what intergalactic space junk exists out there and once it gets there we don't know if, like our solar system, we will find moons, comets, asteroids or rings that we need to avoid. It is not just travel time that is an issue for people and machines but more importantly real time communication.