Cheap Astronomy - Dear CA #138: In Space!

0:00It's the 365 Days of Astronomy podcast, coming in 3, 2, 1.

0:20Hi, this is Steve Nerlick. Why, why, why, why, why, why, why, why Cheap Astronomy? Yeah, why? And this is Dear Cheap Astronomy, episode 138, In Space. There is always a sense of excitement and wonder about putting things in space, either actually, or conceptually, or even imaginatively.

0:50Here at Cheap Astronomy, we do at least do a lot of conceptual and imaginative things in space. For example,

1:01Dear Cheap Astronomy, What is the role of AI in space? Artificial intelligence, AI, simulates human decision-making and finds most practical usage in the sorting and analysing of various kinds of data. So you can feed an AI program pictures of galaxies and the classifications that they fit into, and from there the AI can then identify and classify new galaxies that you feed it.

1:32It's kind of like how Galaxy Zoo used to work, except with Galaxy Zoo, it was people, that is, crowdsourced people, that did the work. These days you can also chat to an AI entity, which will consider what you say and then provide an appropriate response. From the AI's perspective, whatever you say is just data, since talking or texting is just another method of data transfer. And you can also tell AI platforms to draw pictures or compose music,

2:06Although there, you are engaging with an AI platform that has pre-programmed drawing and composing capabilities, and all you're really doing is providing it with parameters to produce an output. In any of these contexts discussed, a degree of machine learning is possible, meaning the more that the AI does, the better it gets at what it does. And if you give it feedback on what it did well versus what it didn't do so well, it will get a lot better at what it does.

2:37That's not the end of the AI story, but it does give some scope to what's possible. And not surprisingly, there is a great deal of scope for the use of AI in space exploration, where you can pretty much use AI wherever data analysis and a protocol set of responses to that data might be required. One can get philosophical here and question how much intelligence there really is in following a set of protocol responses that have been pre-programmed,

3:08but then consider how much of astronaut training involves developing protocol responses. Hopefully every astronaut knows that if there's an air leak, they should immediately do certain responses in a stepwise order, unless the yellow light starts blinking, in which case they should do step three before step two. Again, the whole point of AI is to simulate human intelligence and to mimic human intelligent action. And on the bright side, AI won't panic, get overwhelmed or get tired.

3:41It's reasonable to question AI's capacity for creative thought, but again, even where senior command decisions on a spacecraft may not be entirely protocol driven, they will still mostly involve weighing up the balance of probabilities of going one way or the other based on whatever data is available and the commander's own experience. All that said though, AI isn't ready to take over command of a spacecraft just yet. AI entities are developed for very particular contexts

4:14and it can be challenging to build in capacity for them to respond to unanticipated scenarios, particularly where those scenarios may be unanticipated by an AI's human programmer. There is a role for machine learning there, although learning by trial and error isn't really what you want in a spacecraft's operating system. Nonetheless, AI is improving iteratively, which is not to suggest it has infinite possibility, but it does suggest we are yet to see just how effective it might become.

4:49Hello Steve. Oh hi Bridget. Have you been talking to chatbots? Oh, uh, it's for the podcast Bridget. Really? I've been researching the use of AI in space. What did you discover? Well, it's out there Bridget. Oh yes, you do realise Steve that AI is not a new thing. Any form of autonomous navigation is essentially AI. The Mars rovers launched in 2003 had AI to help them move around on Mars. AI has also been used to analyse astronomical data,

5:22for example to identify exoplanets. Gosh Bridget, you seem to be suggesting the current AI revolution has been hyped up a bit. There's been iterative development in artificial intelligence for well over two decades. So, there you go folks. AI is already in space, and will very likely continue to be so for perpetuity. We may just rebrand it a few times, so it keeps looking like it's something new. We do still have a long way to go before we put HAL 9000 in space,

6:01and plenty of time to think carefully about how exactly we do that. Space is often seen as virgin territory, where we might make a fresh start, and leave behind all the old squabbles and tensions. Indeed, we may have no choice.

6:22Dear Cheap Astronomy, what will space warfare look like? We've previously discussed on this podcast, that fighting a war from space is probably a bad idea. You can conceivably drop a bomb from orbit, but once you've identified yourself as an enemy, it's pretty easy to take you out from the ground. You may be able to dodge a few missiles at first, but once you're out of manoeuvring fuel, you'll just be stuck in a straight line orbit,

6:53and hence a sitting duck. So, in most respects, it's much easier to attack a ground target with a sub-orbital missile fired from elsewhere on the ground. Ship-to-ship warfare in space also seems a bit impractical. If you're in orbit, and you make an aggressive move towards another ship, it will see you coming long before you're in firing range, and it would be easier to radio ground forces to take you out,

7:24rather than bothering to engage you itself. You could always try a Star Trek We are one big happy fleet attack, but that only really works once, and ground forces will take you out shortly afterwards. There are all kinds of robot attacks to imagine, where being taken out by ground forces afterwards might be less of an issue, but there's no real practical difference between doing that and just firing a missile from the ground. Once you are far enough away from Earth

7:56that ground forces become less relevant, you are then in the utter vastness of space. We can imagine that in the long term there might be space lanes that are used by purpose-built passenger and cargo spacecraft, and to protect them there might be additional armed and more manoeuvrable spacecraft, whose purpose it will be to defend the passenger and cargo craft. On top of all that would be pirates and other aggressive spacecraft, essentially the bad guys.

8:28But whether such scenarios will really happen comes down to economics. What makes space travel different from ocean travel is that in space, all you have to do is to get up to speed, after which you will maintain that speed indefinitely in a vacuum. In ocean travel, you have to keep on expending fuel just to maintain a constant speed, and pushing against resisting ocean and air means that smaller, streamlined craft are faster and more manoeuvrable

9:00than large, lumbering cargo ships. But, in space travel, you expend most of your fuel at the start, and then coast the distance you have to travel with just enough leftover fuel for a few manoeuvres at the other end. So, if a pirate wants to chase down such a ship, it's going to have to deal with the rocket equation, meaning the pirate has to carry more manoeuvring fuel than the cargo ship, and if the pirate's carrying more manoeuvring fuel,

9:30it will have to carry more fuel to manoeuvre that manoeuvring fuel, and so on. So, in space, rather than a big cargo ship being overtaken by smaller, faster attack vessels, the attack's spacecraft will have to be bigger than the spacecraft they are attacking. Cost-benefit-wise, that arrangement may just not work for pirates. If they have to invest more up front, then they can never hope to plunder afterwards. Any weaponry attached to spacecraft

10:01will most likely be for close-range attack and defence. While missiles are very effective long-range weapons, they are a lot of mass to carry on board, particularly if you rarely use them, so it probably makes more sense to launch missiles from ground-based or orbiting installations, where maintenance and upgrades are easy to manage. On that basis, it would make more sense for spacecraft to invest in short-range defence and attack weaponry, where defence weaponry

10:32could be used to take out missiles. Remembering that in space, you are going to see these things coming well ahead of their impact, and hence have ample time to prepare your defences. In the unlikely event you do ever have a close-range sortie with another spacecraft, lasers and other somewhat imaginary energy beam weapons, could be effective to burn out peripheral infrastructure, like heat radiators and solar panels, but such weapons are only really effective up close,

11:03since their energy punch declines quite quickly over distance. So if you want to achieve anything more than peripheral surface damage, what you need is good old guns and bullets to achieve any kind of armour-piercing effect. But hey, who knows, perhaps we'll all be too busy exploring the vast wonders of the universe to worry about old territorial squabbles. Nah, just kidding.

11:33This is the end bit. So, there you go. We may learn to live in harmony simply because we have no other option. Of course, once we colonise other planets, the whole ridiculous business of warfare may just start up again, but of course, it may be the case that all the space exploration will be done by our robot progeny, and they will be making their own artificial decisions on how to best manage it all.

12:04Will warfare still be an option in that scenario? Watch this space. But that's it for another episode of Dear Cheap Astronomy. If you've got a space science question, or you just want to be a stochastic parrot, why not write to cheapastro at gmail.com and we'll offer you some hallucinations. Thanks for listening. Steve Nellick, Cheap Astronomy. You are listening

12:36to the 365 Days of Astronomy Podcast.

12:46Cool.

12:53The 365 Days of Astronomy Podcast is produced by the Planetary Science Institute. Audio post-production is by me, Richard Drumm. Project management is by Aviva Yamani, and hosting is donated by Libsyn.com. This content is released under a Creative Commons attribution, non-commercial 4.0 international license. Please share what you love, but don't sell what's free. This show is made possible thanks to the generous donations

13:24of people like you. Please consider supporting our show on Patreon.com forward slash CosmoQuestX and get access to bonus content. Without your passion and contribution, we won't be able to share the stories and inspire the worlds. We invite you to join our community of storytellers and share your voice with listeners worldwide. As we wrap up today's episode, we're looking forward to unraveling more stories from the universe. With every new discovery

13:54from ground-based and space-based observatories and each milestone in space exploration, we come closer to understanding the cosmos and our place within it. Until next time, let the stars guide your curiosity. Be.

14:11Music Music Music Music valt democrats 家 Waiting