(Spoilers] Ending thoughts

#21DestinPosted 3/21/2013 7:31:06 AM
SKA_ posted...
that is assuming the development of an AI wont be a slow evolution, who knows where an evolution of non-human intelligence will end up.


if it's really slow, we would just mess with it all the time. every time it evolves to something we can't understand we'll scrap it and reel it back in.

the only way for a slow evolution to happen is if you let it separate from the creator for a long period of time. would we really be stupid enough to do that? create an intelligent machine and send it on a one way trip into space? create a machine and not know that it's intelligent and send it into space?

Delta123456789 posted...
Destin made the statement broadly along the lines "if we make it we'll make it think like us", but I think creating an intelligence and making it something we can understand and communicate with are two different things. I'd say it would be a lot easier to make something intelligent than to make some intelligent that also sees the world the way we do.


what basis do you have for the last statement? the start to creating intelligence is to teach it and we can't teach what we don't know. i'm not saying we're going to make them like us because that's what we want to do. we're going to make them like us because that's the only thing we know how to do.

two parents who only speak English can't raise a child to speak Latin.
---
Destin the Valiant
#22Delta123456789Posted 3/21/2013 8:43:31 AM
Destin posted...
Delta123456789 posted...
Destin made the statement broadly along the lines "if we make it we'll make it think like us", but I think creating an intelligence and making it something we can understand and communicate with are two different things. I'd say it would be a lot easier to make something intelligent than to make some intelligent that also sees the world the way we do.


what basis do you have for the last statement? the start to creating intelligence is to teach it and we can't teach what we don't know. i'm not saying we're going to make them like us because that's what we want to do. we're going to make them like us because that's the only thing we know how to do.

two parents who only speak English can't raise a child to speak Latin.


What I had in mind was there are more ways to be intelligent than there are to be intelligent AND comprehensible to humans (each addititional attribute lowers the probability), so any intelligence that came about by random chance or experiment that wasn't carefully aiming to be as "human" as possible is likely to be incomprehensible.

Of course the situation is different if people are trying to emulate humans as closely as possible rather than just trying any old means to create intelligence. One problem there is knowing how closely your thought process and intentions match those of the AI. You'd need some way of comparing the two (say using some kind of mind-reading technology) to make sure the computer doesn't differ in some non-obvious way (e.g. just as a parrot can mimic human speech, an AI might give the appearence of humanity but be very different underneath).

(Apologies if this seems rambly or ill-thought-out. I'll rather tired at the moment and just throwing out whatever springs to mind, which may not be fully relevant or aiming at a particular point).
---
DF: So why can a Marauder roll but a multiplayer Turian can't?
Delta1-9: You are fighting female turian husks. You have reach but they have flexibility.
#23Delta123456789Posted 3/21/2013 10:52:06 AM
If anyone is interested, there was an interesting articleon a rationality blog about how and why we make the mistake of assuming that an alien mind, such as that of an alien species or AI, would be pretty much like us:
http://lesswrong.com/lw/so/humans_in_funny_suits/
It starts a bit slow but gets interesting further on.
---
DF: So why can a Marauder roll but a multiplayer Turian can't?
Delta1-9: You are fighting female turian husks. You have reach but they have flexibility.
#24maltzsan(Topic Creator)Posted 3/22/2013 9:55:24 AM
Thanks for sharing the article. Great blog, too.
#25maltzsan(Topic Creator)Posted 3/22/2013 2:04:32 PM
Found an argument above:

"Hard-learned knowledge will be inevitably lost by organics. Does the synthetic Green ending justify this? That is, organics advance too slowly, so a half-synthetic body is a greater good for organics, too."

This argument is based on the assumption that half-synthetic organisms will be able to store a much larger amount of data in their collective (?) database similar to those of the geth's. Let's assume this is true.

My personal experience tells me that many conflicts in this world results from people's bias. Bias is formed when judgment is made from incomplete information. Theoretically, a shared database allow the storage of more information, and therefore there will be less bias, leading to less conflict. If what you know is the same as what I know, we should derive identical logical conclusions.

That's assume that all organics share the SAME database. That's very hard to imagine. How does a green light suddenly links all organics' mind? And are less advanced organics affected to? Where is the line drawn? It is like asking "do only human have a soul" all over again. How to we uplink a bacteria into our database? Hm...

My current thought about the Green synthetic ending is supposedly good, but impractical. Mature Sci-Fi novel turns to fairy tales. Or maybe we organics are too stupid to understand the Green light's great design. I get that a lot from church.
#26Delta123456789Posted 3/23/2013 12:21:41 PM
maltzsan posted...
Found an argument above:

"Hard-learned knowledge will be inevitably lost by organics. Does the synthetic Green ending justify this? That is, organics advance too slowly, so a half-synthetic body is a greater good for organics, too."

This argument is based on the assumption that half-synthetic organisms will be able to store a much larger amount of data in their collective (?) database similar to those of the geth's. Let's assume this is true.

My personal experience tells me that many conflicts in this world results from people's bias. Bias is formed when judgment is made from incomplete information. Theoretically, a shared database allow the storage of more information, and therefore there will be less bias, leading to less conflict. If what you know is the same as what I know, we should derive identical logical conclusions.

That's assume that all organics share the SAME database. That's very hard to imagine. How does a green light suddenly links all organics' mind? And are less advanced organics affected to? Where is the line drawn? It is like asking "do only human have a soul" all over again. How to we uplink a bacteria into our database? Hm...

My current thought about the Green synthetic ending is supposedly good, but impractical. Mature Sci-Fi novel turns to fairy tales. Or maybe we organics are too stupid to understand the Green light's great design. I get that a lot from church.


Glad you liked the blog.

My take on the "becoming partly synthetic" thing was that organics gained the ability to communicate with and understand synthetics. I don't think that requires they start sharing information necessarily, or end up some sort of hive mind.

I'm also not sure that becoming superintelligent right away is the best idea. There's a sequence of posts on that Less Wrong blog suggesting it might actually be more fulfilling for each individual to learn and examine the world themselves, allowing them the joy of discovering things rather than just being told "this is what we know". The main poster argues allowing for a gradual increase in intelligence over time would be more enjoyable than immediately making yourself as intelligent as possible.

Obviously that wouldn't extend to things like finding a cure for cancer, some problems (e.g. death and great suffering) are too painful not to be solved right away.
---
DF: So why can a Marauder roll but a multiplayer Turian can't?
Delta1-9: You are fighting female turian husks. You have reach but they have flexibility.
#27VKWANPosted 3/24/2013 9:03:42 PM
My take on the Organic vs Synthetic debate;

Unlike all civilizations after the Reapers were created, the Leviathans chose their own technological path of advancement. Every subsequent cycle was designed by the Reapers to follow a set pathway that ensured the success of the harvest. In fact, I believe that all of the advanced species we've seen are all humanoid and roughly about the same size (compare to the Leviathans, who are HUGE and seem to be marine-life invertebrates) because this is the design of the Leviathans and the Catalyst.

From what the Catalyst indicates, there is a pattern of Organic vs Synthetic (for all we know though, this could very well just be Reapers vs Organics). Yet, it could be argued that it is precisely because of Reaper influence that every cycle progressed to the point of developing Synthetic life that must rebel. Even with our technology today, we've noticed that our style of thinking changes with whatever technology we use. If Organics were left to progress in a way that wasn't affected by the Reapers, they may very well create Synthetics that don't rebel. Maybe disastrous Synthetic beings are an inevitable by-product of the Leviathan's (and only their) technological path of advancement.

But the Leviathans couldn't have been stupid -- they label themselves the apex species, and that's a justified label. They must have had some reason to believe that Organic and Synthetic life are incompatible. Maybe there is something inherent to all Synthetic beings that make them rebel. Maybe that urge to rebel is hardwired in self-aware Synthetics like how Organics are hardwired to try to survive.

Yet, for what it's worth, you can play ME3 and provide a scenario where that isn't true. EDI is very well received on the Normandy. The Geth can be reconciled with the Quarians. Note the Geth chose NOT to eradicate the Quarians when the Quarians fled Rannoch. That seems like a huge detail to me. Then again, Commander Shepard (acknowledged even by the Leviathans) and this cycle in general (thanks in equal parts to the Protheans and Shepard) are pretty much anomalies, so it's still possible (if less credible) for the Catalyst to claim that Organics and Synthetic life are always going to fight.

As I see it, the only explanation is thus: The Leviathans were lazy and apathetic when designing the Catalyst. They carelessly programmed the Catalyst with a bad understanding of the project parameters. What constitutes as preservation of life? Not specified, so the Catalyst thought it was okay to turn things into Reapers. What variables should the Catalyst seek to influence? Not specified, so the Catalyst messes with everything it can. The programming here was sloppy, and so the cycle began inadvertently. Bad programs with vague, undefined parameters is a case of "The Created will always rebel against the Creators" that anyone with any experience in Computer Science can relate to.

Other ways we can tell that the Catalyst is just a bad program:

1) Infinite Do/While loop
2) Program does not EVER terminate unless external device is inputted (...what?)
3) Does not have a sustainable solution (I'd imagine planets were uninhabitable after particularly massive Reaper invasions. Life is a finite and rare thing, so in my mind, the Reapers probably destroyed a lot more than they preserved.)
4) I'm actually just a novice programmer but I'm sure better programmers are able to see the parallels here.

That's my biggest problem with the story -- The plot exists in this universe because Leviathan programmers don't know how to code properly. The Reapers can hack Geth, why can't they just hack all Synthetic beings to become peaceable to Organics? (One explanation is that the Reapers don't know how to do that, which would make them seem kind of stupid). That's just ONE WAY they could've dealt with this supposed inevitable conflict. Instead, the Catalyst decided on this convoluted mess of a plan...
#28Delta123456789Posted 3/25/2013 2:57:12 AM
You raise an interesting point about how Reaper influence may have perpetuated the cycle by making civilisations and their technology less varied, but as you say the problem existed before the reapers did (so it was at least a problem in the Leviathans' time).

As for whether AIs are "hard-coded" to rebel I don't think that's the case in the real world. Humans' self-preservation instinct is the result of natural selection (the ones who struggled to survive had kids, the others didn't), but AI is created by a completely different process so wouldn't necessarily have the same characteristics as humans and other animals.

I agree the creator of the Reaper AI was an idiot. "How shall we deal with AI rebellions? Let's build a super-powerful AI with vaguely-defined role!" Then again the AI did imply it tried lots of other things before the Reapers and the cycle kept happening, so maybe it started out okay (trying minor influences etc.) until it decided only extreme measures would work (though the fact it was allowed to consider extreme measures in the first place is a massive design flaw).
---
DF: So why can a Marauder roll but a multiplayer Turian can't?
Delta1-9: You are fighting female turian husks. You have reach but they have flexibility.
#29maltzsan(Topic Creator)Posted 3/26/2013 9:49:22 AM
If organics can be understood as biochemical machines, we are not fundamentally different from synthetics. Whatever happened to us could happen to the "synthetics".

Would AI "natural select" to gain a hardwired trait of self-preservation and altruism? AIs who do not value self-preservation would have a higher chance to be terminated, either by the wary organics or other AIs. AIs who do not value others would be outcast by the AI society. Eventually, AIs may be just as selfish and nice as organics.

I also noticed an interesting detail of the game. When Javik came on board, the Alliance protocol was "assume hostility by default". This is how conflict arise in organics - unknown brings fear, which triggers our adrenaline to give a Fight or Flight response.

However, for Geth, their unknown brings uncertainty, and a lack of action to terminate the quarians, for they cannot predict the outcome of terminating an entire species. They lack "adrenaline" and the often irrational fight or flight instinct.

Maybe irrationality is the fundamental differences between the organics and synthetics. Some of the ideas above lament the stupidity of synthetics. But maybe it is the organics who are the sources of all evil, and synthetics simply lost faith in organics (until they meet Shepard if certain plot elements fulfilled).

National Selection gave us this tendency to burn witches and accuse others for anything that does not fit our tiny mind. I still see that happening every day, right here, right now.

Catalyst and the Reapers in a sense is the "I give up" option on organics. The Green ending sounds equally hopeless.
#30Delta123456789Posted 3/26/2013 10:32:00 AM
maltzsan posted...
If organics can be understood as biochemical machines, we are not fundamentally different from synthetics. Whatever happened to us could happen to the "synthetics".

Would AI "natural select" to gain a hardwired trait of self-preservation and altruism? AIs who do not value self-preservation would have a higher chance to be terminated, either by the wary organics or other AIs. AIs who do not value others would be outcast by the AI society. Eventually, AIs may be just as selfish and nice as organics.


An AI having human traits like the desire to procreate (create other AIs) or a need to socialise or tendency to be corrupted by power would be like it having red blood. There's just no reason to design one that way, and even if an AI appeared randomly without deliberate creation (which seems much less likely in reality than fiction) it wouldn't have have these characteristics because it hasn't been subject to selection yet.

I recommend reading up a bit on natural selection just because it's interesting. This series of blog posts is a good start if you are interested:
http://wiki.lesswrong.com/wiki/Evolution#Blog_posts_.28sequence.29
What surprised me is how slow and stupid natural selection is, and how it doesn't care about species but individuals (i.e. it won't produce an individual that gives up its own breeding opportunities for the sake of the species, it only favours traits that make an individual's own genes more likely to reproduce).
---
DF: So why can a Marauder roll but a multiplayer Turian can't?
Delta1-9: You are fighting female turian husks. You have reach but they have flexibility.