Jump to content

NASA mandated to do manned mission to Mars


NFUN

Recommended Posts

1 minute ago, DerekL1963 said:

No, if you actually read the bill rather than the breathless and overhyped interpretations of it...  Congress mandates that NASA produce a report on how we might go to Mars in the next 25 years, and reminds them in no uncertain terms that the Senate Launch System is the preferred way to go about it.

Yeah, I was afraid it was something like that. I was in a hurry to do something so I didn't read it over. Apologies.

Link to comment
Share on other sites

Haven't they already done countless reports on how to go to Mars ?

https://www.nasa.gov/pdf/373665main_NASA-SP-2009-566.pdf

The last one was a ridiculous 11 SLS launch architecture with ISRU refueling on Phobos.

https://www.nasaspaceflight.com/2015/09/sls-manifest-phobos-mars-2039/

Edited by Nibb31
Link to comment
Share on other sites

2 hours ago, Nibb31 said:

The last one was a ridiculous 11 SLS launch architecture with ISRU refueling on Phobos.

https://www.nasaspaceflight.com/2015/09/sls-manifest-phobos-mars-2039/

Nibb, have you discussed this in depth anywhere else or have a link to an in depth discussion of this architecture?  I'm rather curious to hear what NASA did wrong or how they might do it better.  

I mean, I take it that doing 30 falcon heavy launches instead would let you skip ISRU and probably save money vs SLS launches?

Link to comment
Share on other sites

5 hours ago, SomeGuy123 said:

Nibb, have you discussed this in depth anywhere else or have a link to an in depth discussion of this architecture?  I'm rather curious to hear what NASA did wrong or how they might do it better. 

The discussion for that proposal is here:

http://forum.nasaspaceflight.com/index.php?topic=38487.0

Link to comment
Share on other sites

I wish they had done something sensible like mandate undertaking the initial phases to begin setting up orbital infrastructure (followed by lunar)  to provide the foundation to begin planning a mission to Mars . . . 

We do not even have a space dock or fuel depot up there yet, much less an orbital construction yard. As far as I know we have done ZERO to explore the prospect of extracting resources from Luna; and yet we are gonna go for Mars? You gotta slither before you crawl . . .

Sounds like both parties are trying to capitalize on the current hype and improve their PR a bit.

Link to comment
Share on other sites

17 minutes ago, Diche Bach said:

I wish they had done something sensible like mandate undertaking the initial phases to begin setting up orbital infrastructure (followed by lunar)  to provide the foundation to begin planning a mission to Mars . . . 

We do not even have a space dock or fuel depot up there yet, much less an orbital construction yard. As far as I know we have done ZERO to explore the prospect of extracting resources from Luna; and yet we are gonna go for Mars? You gotta slither before you crawl . . .

Sounds like both parties are trying to capitalize on the current hype and improve their PR a bit.

You can do a Mars mission with a few SLS launches, case and point the Boeing Affordable Mars Mission.

One launch delivers a cargo lander/surface hab and an SEP tug to get it there, the SEP tug comes back, another launch delivers the MEM and TransHab, and another launch delivers an Orion and a kick stage.

Link to comment
Share on other sites

28 minutes ago, _Augustus_ said:

You can do a Mars mission with a few SLS launches, case and point the Boeing Affordable Mars Mission.

One launch delivers a cargo lander/surface hab and an SEP tug to get it there, the SEP tug comes back, another launch delivers the MEM and TransHab, and another launch delivers an Orion and a kick stage.

Just because you "can" do something does not mean it is the best use of resources. In the 1970s, they launched two spacecraft that are now either beyond or nearly beyond the solar system. But neither of those missions have truly brought us any closer to routine interstellar space travel. If the project design does not substantially improve the safety, efficiency, affordability or general ease (meaning likelihood of becoming 'routine') of the mission type, it is a 'flash in the pan.'

We don't need to get to Mars nearly as badly as we need to make real progress toward being a space faring species. There are resources to be had out there which can have dramatic positive effects on humanity. IMO, simply "going to Mars" does less to bring those resources within our grasp than more superficially mundane efforts that establish the off-Earth infrastructure to make extra-planetary missions so safe, efficient, affordable and easy that they become mundane.

Link to comment
Share on other sites

1 minute ago, Kerbart said:

The workings of the US Government remain a mystery to me. Can't the next president simply cancel these plans, like Bush's return to the moon campaign was cancelled? Or would it take an Act Of Congress this time to do that?

It is posturing. Of course is can be 'cancelled' even if it cannot be "CANCELLED."

Link to comment
Share on other sites

4 hours ago, Diche Bach said:

There are resources to be had out there which can have dramatic positive effects on humanity. IMO, simply "going to Mars" does less to bring those resources within our grasp than more superficially mundane efforts that establish the off-Earth infrastructure to make extra-planetary missions so safe, efficient, affordable and easy that they become mundane.

You know, we don't actually need to do that, either.  The solution to this problem isn't to collect resources from asteroids.  Not yet.  The solution is to develop either some form of self replicating factory (either via a bunch of really compact and flexible robots or nanotechnology) or some form of useful AI (by improving tensorflow and studying the synaptic mapping of the human brain and developing simpler scaffolding AIs that can read a synaptic map and extract the pattern of how the brain's regions are organized).

If we were a rational society we'd spend all 18 billion that would go to NASA, and maybe another 200 billion that would go to the military, and maybe another few hundred billion that would go to medicare (we'd start freezing hopeless medicare patients instead of wasting money to extend their lives a few weeks) into this effort.  Maybe we'd even tax heavily our richest members.  Do we as a society need more mansions, or should we spend instead spend a couple trillion a year developing the technology to defeat our species's biggest problems?

Anyways, obviously with some form of AI, some form of self replicating factory, some form of digital immortality (a form of AI but it would mean our society would have members who have a longer view than the next 30 years) - with some or all of things like that, most of the problems of space would plummet in difficulty and become almost trivial.  And all that stuff can be developed on earth, using massively scaled up versions of what we already have.  Don't see any reason we need metric tons of platinum and rare metals to do it, either...

Edited by SomeGuy123
Link to comment
Share on other sites

Well, before we get too carried away: define intelligence, much less "artificial intelligence."

The term "AI" gets used for everything these days, any algorithm-based mechanism which exhibits some degree of purposeful behavioral reactivity. As far as I can tell, what is at the cutting edge of "artificial intelligence" is more like "sophisticated algorithms" and the chasm between replicating animal, much less human behavioral flexibility and adaptiveness is still very large. I can say with some confidence that the distances between the various sub-disciplines of behavioral sciences which will prove to be integral to a coherent model of "intelligence" are presently sufficiently disconnected and theoretically remote that--even if we had handheld fMRIs and the ability to do intensive longitudinal research on sufficient numbers of individuals over whole lifespans so as to establish how learning works--we would still be several generations away from valid models of "intelligence."

Link to comment
Share on other sites

23 hours ago, DerekL1963 said:

Congress mandates that NASA produce a report on how we might go to Mars in the next 25 years

I thought they were supposed to make use of SLS and not GMAT and Powerpoint.

6 hours ago, Kerbart said:

Can't the next president simply cancel these plans, like Bush's return to the moon campaign was cancelled? Or would it take an Act Of Congress this time to do that?

Given that the bill only outlines the budget for 2017, including a measly $1.4 billion for yet another series of mostly conceptual studies, one can simply fail to mention anything long these lines in the 2018 budget.

And even this budget can get blocked by Obama.

Link to comment
Share on other sites

14 hours ago, Diche Bach said:

. I can say with some confidence that the distances between the various sub-disciplines of behavioral sciences which will prove to be integral to a coherent model of "intelligence" are presently sufficiently disconnected and theoretically remote that--even if we had handheld fMRIs and the ability to do intensive longitudinal research on sufficient numbers of individuals over whole lifespans so as to establish how learning works--we would still be several generations away from valid models of "intelligence."

Define generations*.  Also, we wouldn't use fMRIs.   Too low resolution.  The approach that might work to get there would be first to make connectome or synaptome.  (the mapping of every single connection in a preserved brain or even the individual synapse strengths - a big huge project that would require billions of dollars and warehouses full of slicing machines and multiple electron beam microscopes)

Then we'd develop what I called a "scaffolding AI".  This thing would be able to look at the synaptome and connectome and essentially convert it from a morass of billions of connections to a color coded visual map based upon the patterns common to each area.  Then you'd build other AIs that essentially are a program, probably written in tensorflow, that can take a region from this brain map and write a program in tensorflow that mimics the connectivity patterns of that region.

Then you'd spend many years developing ever more sophisticated systems that have ever more complex internal models and internal representations of the world.  Having stolen heavily from the human brain your overall machine would be vaguely human like (eventually the machine would have many internal memory regions where it has a concept of it's own relationship to a model of the world, where it models possible courses of action against that model, and so forth).  But yeah, if alphaGo is a few dozen systems working together, this would be thousands of them.  A vastly larger and more complex system.  You'd need moar simple AIs just to detect failures and help you debug this thing...

Nevertheless the basic concept is to take what we have - simple but powerful tools that use a form of artificial neural network to be able to perform many well defined tasks better than humans - and to use it to build the scaffolding that would eventually let us make a fully sentient AI.  The fully sentient AI would be a massive machine composed of thousands of artificial neural networks, with rules and connection patterns borrowed from a once functioning brain.  Sort of how before people could create Python they needed to write assemblers and then Pascal/C programming languages in order to write Python in.  And create a whole support library of algorithms and an OS to provide an environment so it would run on multiple computers.

Once you realize the scale of the task - thousands of interdependent systems each as complex and hardware hungry as alphaGo - you realize that various people who thought they could invent AI decades ago were smoking the good stuff.  They had no inkling of the scale of the task.  That's why there is a false perception that the problem is intractable.

That's one approach.  But even a far simpler one where you start with a defined set of tasks would be useful.

For example, a machine that uses a kinematic model (similar to an autonomous car) and can manipulate components in a factory.  So it would internally know the shape and weight distribution of components.  It would be trained to recognize said components reliably with just a few reference photographs.  It would be able to pick the components up and manipulate them to the final state in a mechanical design for a part.

So essentially it would let you design a machine in a CAD program and the factory robots would be able to build most designs with minimal training.  (just like a human, it might need some hinting as to correct assembly order)  Not as sexy as Hal9000 but achievable in the near term.  You could build robots able to repair other robots, so long as the failed component is in an isolated, replaceable module with a simple type of attachment, by swapping parts.

I'm saying if we had a rational society we'd dump the lion's share of non essential GDP into things like this.  


* In conclusion, true general artificial intelligence is closer than "generations" if you meant 60-90 years, because the road to getting there is to combine various advanced general tools like tensorflow and recent advances in brain scanning technologies to speed up the process of getting there.  That process is an exponential one, it might seem like we're only 10% of the way to AI when it's only 10 years away as a result.  As for the definition of it : all it really means is that any useful task a human can do, it will be possible to use a collection of AI tools to make an AI that does the same thing at at least average human levels of competency.  Not one omni-being that has emotions and a voice and all that jazz.  

Edited by SomeGuy123
Link to comment
Share on other sites

So you think that making maps of the connectivity of neurons of cadavers is going to be the key to making AI? Right . . . cause all you need is a map of a dead brain and you understand all about how it functioned while it was alive . . .

Neat stuff, just about as neat as the stuff in a hundred other disarticulated sub-sub-sub-fields in the suite of salient disciplines, but I fail to comprehend how maps of neural connectivity will somehow magically create a definition of intelligence, much less artificial intelligence, much less a method to "create" said artificial intelligence. Seriously, if you cannot provide a coherent definition of intelligence, then why are you even speculating? Does google have a definition of "intelligence?" If not, they they are _still_ smoking the good stuff, because to set out to engineer something you cannot define is silly.

Without a precise definition of intelligence, then all you are doing is arm-waving: "Look at how much more optimized/efficient/accurate/taxonomicallycorrect this process is! We made an advance toward intelligence!" Maybe, maybe not. If you don't actually know where you are going, then making a few steps in any given direction toward "order" and away from "entropy/chaos/randomnicity" does not necessarily constitute advance toward artificial intelligence.

I've spent 25 years studying around the margins of the subject area (the psyche), with some devoted forays into the heartland and I cannot define "intelligence," at least not in a way that is universally valid, i.e., across species, across individuals, across contexts, across developmental phases, etc. If Google and the rest of the big boys who are playing in this area cannot even manage to try to do it properly from the start: define the natural bio-behavioral phenomenon they are seeking to replicate in computer systems, then in my opinion all their fancy techno babble is just that . . . although it is true that if you get enough monkeys typing on keyboards for long enough you will eventually compose all the works of Shakespeare . . .

Link to comment
Share on other sites

9 hours ago, SomeGuy123 said:

any useful task a human can do, it will be possible to use a collection of AI tools to make an AI that does the same thing at at least average human levels of competency.  Not one omni-being that has emotions and a voice and all that jazz.  

"any useful task" I really don't think you mean that because:

smearing peanut butter on a piece of bread

developing an hypothesis

kicking a football

writing a book report

performing an above average competency poetry reading

training a dog

duck hunting

prospecting for oil

planning a mission to Jupiter

designing a new rifle cartridge

painting a nice portrait

trolling for LOLs on the interwebz

making funny memes

doing a game review that players find useful/funny/compelling

designing a new operating system (or even just tweaking an old one)

making art assets for a computer application

packing up for moving to a new state

making breakfast

etc., etc., etc.

Eventually? Sure, I can believe that eventually we will have machines that can do (or at least serve as helpers in doing) any and all these things. Even then, I'm not sure I'd be willing to accept that a robot that could do even one of these things was "intelligent." Functionally competent in a narrow spectrum of tasks? Even to some extent creative or improvisational? Resourceful? Efficient? Problem-focused? Sure, I can believe all that will be doable and maybe even in the next generation or two. But I would not consider that alone to be intelligence.

When computers can do things like: create the next Minecraft, or become the next Warren Buffet, or get itself elected President, then we will be SURE that they have created "artifical intelligence" but short of those sorts of lofty thresholds of human accomplishments, and moreover, the whole suite of attached characteristics (Warren Buffet is after all, not merely a dude who got rich on the stock market, he is also a father, a grandfather, a role model, a community member, a commentator, a writer, a leader, a WHOLE PERSON . . .) I'll have to politely refuse to sincerely refer to such things as "AI" (though I will continue to do so as a shorthand convenience since everyone else on the planets uses the term to refer to even the "simplest" decision-making algorithsm such as those that allow computers to be competitive chess players). 

Link to comment
Share on other sites

other things I'd like to see a computer do:

recognize itself in a mirror

lie to me because it felt it needed to (not simply because the guy/gal who programmed it told it it needed to)

make a termite stick and fish for termites at least as efficiently as a 6 year old Chimpanzee pup

play fetch at least as coherently and devotedly as my 2 year old dachsund

tell me a truly funny and original joke

make an Acheulean hand axe, even just an Oldwan chopper

decide on a good campsite

build a hut out of materials in the natural environment

flirt with a "conspecific"

experience frustration

know love in all its irrational power and glory

sacrifice itself for a greater good

fear death and contemplate the after life

grieve

speculate about the origin of everything

Edited by Diche Bach
Link to comment
Share on other sites

This thread is quite old. Please consider starting a new thread rather than reviving this one.

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...