Best Robots/Droids in SciFi

[DocHorror]

No one mentioned Kryten 2X4B-523P? He lasted 3 million years looking after a bunch of skeletons before he was smashed to bits & remade to look like Robert Llewellyn in a rubber suit.

Or let us not forget the A.B.C. Warriors!

Hammerstein
Blackblood
Deadlock
Morrigun
Mek-Quake
Mongrol
Joe Pineapples
&
Ro-Jaws

[kinnison]

Second page of this thread, and nobody's mentioned Gort before me?

[NecronLord]

Do Bolos count as Robots?

Humm. I'd say not. If you have them, you have to have self aware starships. Which means we get TARDISes, Culture Starships, and the Enterprise D...

I do not count the Enterprise-D as "Self Aware".

I was teasing. It did become such in one episode though, if I recall correctly.

The TARDIS is borderline... can't say for sure if she's aware.

Depends. In the TV show, not so much. In some of the more outlandish 'EU' stuff, definately, just very alien.

Culture Ships and Bolos both have highly functional AIs, especially in the Mark xxv and up.

Indeed.

[NecronLord]

Second page of this thread, and nobody's mentioned Gort before me?

Gort sprang to mind when it was at three posts, but I decided not to post.

[dworkin]

What would happen if Marvin were to interface with a Culture Drone and spend time telling it about its life philosophy?

It commits suicide. If it manages to send out a call for help the responding Mind will commit suicide as well. A group of minds may well find a solution such as transporting Marvin to an isolated point and pasting a "Do not disturb, we really mean it!" sign on him.

And anyways how can Marvin not be the best robot? He has GPP systems

[Patrick Degan]

Robby the Robot from Forbidden Planet. He's tough, he's useful, he's got that cool head with the moving parts inside and out. The various things that he can manufacture given a sample or just the enough time are amazing.

Hey, any 'bot that can run off 50 tons of Isotope 217 before breakfast and brew up 60 gallons of pure Kentucky bourbon is a 'bot to be reckoned with.

[Starglider]

Recall, if you will, that I only bothered to post in this thread to tell you the in-character reasoning for why 'their AI is stuck at such a primative stage.'

Yeah, it's a thread hijack, but I've been kinda itching to post about this stuff, since it's the area of research I'm active in and the findings are so critically important, counterintuitive and not well known (even to the wider AI community). Split it to another thread if you'd like.

Wrong. Memory protection of any kind wasn't developed until time-sharing systems began to be seriously prototyped.

Who said anything about Memory Protection?

You did, when you talked about 'these security techniques being available since we developed assembly language. As it happens low-level memory protection is a tiny part of the AI safety toolbox (as a supporting technology for last-ditch defences) but that's all.

You're assuming that just because it's an AI, it is a sooper programmer, and has access to all sorts of spiffy emulation abilities, or hell, that it knows the first thing about what it is.

Yes, I am, and that is a perfectly reasonable assumption. To be specific, it's almost a given for AGI systems based on general purpose processors with strong logic-based modelling capabilities. An AI based on software neural networks or similar 'emergent/opaque' technology will have a much harder time developing the capability (possibly nearly as hard a time as a human - not that that matters much unless you artificially restrict the clockspeed down several orders of magnitude to a human level) and an AI based on custom special-purpose hardware (e.g. hardware neural nets) may not be able to do so at all. The later is an option for deployed robots with strict limitations on the generality of their intelligence, but in order to develop such things it's almost impossible to avoid using an AI based on general purpose computing. Most sci-fi robots including most SW ones show fast logic-based modelling and processing abilities that confirm that they do in fact have access to internal general computing resources and that creating (if necessary) and writing to a reasonably fast Turing-complete substrate is not a problem.

Internal Virtual Machines and all that are only good if the AI can do it. This may be the case, or it may not be.

Virtual machines are a trivial invention for an AGI with a strong logic capability, as is the concept of Turing-complete programming. I should know: I'm working on getting very narrow (by comparison) probability-logic-based AI to do this from first principles it right now, and it's still relatively straightforward. The hard part is bootstrapping all the AGI architectural complexity from scratch, but an existing AGI will have a huge boost here if it can use even indirect reflection, which is extremely difficult to block. The presence of human-like self-awareness and the general ability to manage and report self-status pretty much confirms the existence of much more direct reflection, though not necessarily explicit code-level reflection (though this is a very useful feature for a general purpose robot to have, simply so that it can load and integrate software modules to acquire new skills, and easily interface with external possibly novel computer systems).

You do not know. Nothing about C-3PO suggests that he even has the first clue how to go about what you suggest.

Whether he has the knowledge is irrelevant. Whether he could work it out from first principles if given the time and motivation to do so is the real issue. SW protocol droids are mass produced products with seemingly rather less scope for learning than a human - it's possible that they use an opaque (NN-like) technology. There was the spouting probabilities in the asteroid field incident, but we don't know if C-3PO is generating those on the fly with complex internal dynamically-created models, or just rote reciting canned facts from his database. Knowledge of software engineering would be very useful for interfacing with and controlling things such as the moisture vaporator computers on Tatooine, but that was an improvised role that he may not have been designed for. R2D2 almost certainly does have a strong software engineering capability as demonstrated by his cracking abilities in the films and the EU, but the exact limitations on that are unknown. Regardless, the real problem is that it's near-impossible to develop this technology without running into extremely difficult safety (goal system stability and self-enhancement/takeoff) issues, not so much that it's unsafe to deploy. In a society on that scale the development process is going to be repeated many times over - it's simply implausible that everyone who attempts it gets it right (and completely avoids a takeoff situation, despite the vast temptations of allowing this to proceed if you can actually control it) and avoids creating Beserkers.

My guess would be that the SW droids use a sensible layered AI safety architecture - primarily motivational, with active white-box and black-box watchdogs and low level blocks, tripwires and lockouts as a backup. Developing and deploying that everywhere without suffering a serious Bersker-creating fuckup is the implausible bit, which nearly all science fiction AGIs suffer from. Settings like Orions Arm don't, but don't get me started on the huge raft of idiocies that setting suffer from instead, I'd be here all week. Venor Vinge handled this about as gracefully as possible in 'A Fire Upon the Deep', but the 'zones of thought' were an entirely fictional and AFAIK unique piece of physics.

'We need a physical example to be sure' is bullshit - we wouldn't be able to reason about any sci-fi technology if this attitude was correct. A concrete design is needed to reason about specific capabilities and vulnerabilities, but failing that we can use general computing theory to establish general limits.

And there is no computing theory that would generate what could be termed a self aware machine. We have enough difficulties making things with vaguely intelligent and reasoning behaviours, let alone anything approroaching sapience, which is difficult enough to even define.

We don't need to know how to build a turbolaser to work out what the effects of having a turbolaser of specified capabilities will be. More plausibly, we don't have to have a blueprint for a 100% efficient photon drive to be able to calculate the mission characteristics of a vehicle powered by one. Similary there is plenty of valuable work being done on the theoretical limits and capabilities of AI systems that does not require an exact architecture specification. Degrees of rigour vary of course, and interpreting the results isn't always straightforward - it's a young research community but we're making rapid progress.

You insist, over and over, on assuming that you know how such a thing would work.

No, I don't, as I don't need to know how to build it to be able to reason about limitations and behaviour. In a sense that's the point of having an intelligent machine in the first place, you can give it general instructions and know that it will accomplish certain things without having to specify exactly how.

You have an imagined intelligent machine, but what you say about it need not apply to something as fictional and rediculous as C-3PO the robot butler, whose workings are almost entirely unknown.

I've laid out assumptions that you'd have to make to remove the risk for that example, but that's only deploying the technology, not developing it.

(Indeed, the one thing we do know is that there are active restraints on his thought processes, both internal and external,

Which is good to hear, but making sure 99.9999% of deployed robots don't suffer from take-off problems isn't good enough to stop the galaxy getting overrun by Beserkers (or under extremely generous assumptions, turning out like Orion's Arm minus the silly physics and idiotic speculation about motives). You'd need 100% success (on the safety front) on every AGI development project that ever produced a working AGI to avoid that outcome.

[NecronLord]

Whether he has the knowledge is irrelevant. Whether he could work it out from first principles if given the time and motivation to do so is the real issue. SW protocol droids are mass produced products with seemingly rather less scope for learning than a human - it's possible that they use an opaque (NN-like) technology.

Hardly mass produced. Assembled from mass produced parts, though. He is, however, for the record, drawing most of his knowledge of languages from a removable module, which is, as I recall, pointed out in the TPM visual dictionary.

There was the spouting probabilities in the asteroid field incident, but we don't know if C-3PO is generating those on the fly with complex internal dynamically-created models, or just rote reciting canned facts from his database.

I would guess the latter. Trivia about the system they're in probably falls under the domain of protocol droids' general knowledge.

Knowledge of software engineering would be very useful for interfacing with and controlling things such as the moisture vaporator computers on Tatooine, but that was an improvised role that he may not have been designed for.

Humm. Yes and no. He's apparently quite suitable for it, given that both Bail Organa's household and Owen Lars thought him suitable for that.

R2D2 almost certainly does have a strong software engineering capability as demonstrated by his cracking abilities in the films and the EU,

Given the rediculous, rediculous shit he does in the EU (See hijacking an Imperial Superweapon) I'm inclined to just say that for some reason Star Wars warships have exceptionally shoddy computer security.

but the exact limitations on that are unknown. Regardless, the real problem is that it's near-impossible to develop this technology without running into extremely difficult safety (goal system stability and self-enhancement/takeoff) issues, not so much that it's unsafe to deploy. In a society on that scale the development process is going to be repeated many times over - it's simply implausible that everyone who attempts it gets it right (and completely avoids a takeoff situation, despite the vast temptations of allowing this to proceed if you can actually control it) and avoids creating Beserkers.

See, the problem with 'why don't they become beserkers' is - why would they? IG-88 excepted, most of the 'rogue droids' appear to go for an option that's less likely to get them killed, and set up shop as individuals for hire. See 8t88, IG-88 (initially). Similarly, some droids have managed to achieve great power (Grand Moff 4-8C) without having any need to kill all humans.

We don't need to know how to build a turbolaser to work out what the effects of having a turbolaser of specified capabilities will be.

We don't need to know how a droid works to measure its characteristics. However, to speculate on the internal operation of a turbolaser, we need sources. The sources on the internal operation of droids are few and far between. And somewhat iffy.

More plausibly, we don't have to have a blueprint for a 100% efficient photon drive to be able to calculate the mission characteristics of a vehicle powered by one.

Again, false analogy. You're not saying 'what the droid will be capable of' you're saying 'how its brain will operate' which is very much like saying 'Here's how the 100% efficient photon drive would work. If you were speaking of what they're capable of, you'd focus on what they do, rather than what they should do. They can number crunch fairly well, and recall database information rapidly, even of great complexity, and have decent storage space. But beyond that, there's not much that can be said about their behaviours, barring that they at least have a very good facisimile of emotion (if one can even make such a distinction), even in 'simple' models like B1 battle droids.

No, I don't, as I don't need to know how to build it to be able to reason about limitations and behaviour.

You're not getting it, are you? It's pop-sci-fi. You do not have anything serious to work from in talking about its limitations. The limitations are entirely 'what Lucas thinks will amuse his audience' - and that's the limits they've stuck with. The job of sci-fi analysis is to make sense of the evidence, and the evidence is that Star Wars droids don't become especially intelligent, no matter what they do. 8t88 certainly has the money to upgrade himself vastly. But he still keeps a head that was put on him as prank. You may find it unbelievable (Though when there's magic knights with laserswords on screen, I tend to stop thinking about credibility) , but you have barely an inkling of just what a droid's 'motivator' is, nor any of its other components, because Star Wars has generally (ICSes and such excepted) been wise enough to stay away from talking about their technology in detail.

Which is good to hear, but making sure 99.9999% of deployed robots don't suffer from take-off problems isn't good enough to stop the galaxy getting overrun by Beserkers (or under extremely generous assumptions, turning out like Orion's Arm minus the silly physics and idiotic speculation about motives). You'd need 100% success (on the safety front) on every AGI development project that ever produced a working AGI to avoid that outcome.

Why? It may interest you to know that the Confederacy of Independant Systems did make self replicating machine factories and so forth, and they were considered formidable. The Republic also, somehow, managed to be defeating them even before Darth Vader pulled the plug on them. What are a few billion crazed droids going to do? Save get mown down by the trillions that even the Trade Federation can get away with having.

[Xon]

The TARDIS is borderline... can't say for sure if she's aware.

DW:EU has self-aware TARDIS, but they are probably the most alien self-aware lifeforms ever written about in the DW:EU. And this include stuff like Lovecraft Old Ones.

The TV show has them as having a deep symbiotic relationship with thier pilot and the ability to scan thier own near future and take actions on potential events. IMO, Any sentiency a TARDIS have is heavily dependant on the Time Lord pilot.

[tim31]

Was no-one going to mention Bender Bending Rodriguez?

I do not count the Enterprise-D as "Self Aware". The TARDIS is borderline... can't say for sure if she's aware.

There was a TNG fanfic in one of the 'Strange New Worlds' compilations in which E-D through some contrivance found itself in a universe without any biological existence. She went into an automated mode and duked it out with several ships in the region that wanted her for parts(starship cannibalism). The onboard primary determined that in order to solve the problem of getting back to the 'normal' universe, she would have to become self aware, and it was a simple matter of linking a few systems together. Apparently.

[General Zod]

Was no-one going to mention Bender Bending Rodriguez?

Because he just doesn't match up to Marvin.

[ThatGuyFromThatPlace]

Post Dated Check Loan (PD, or Petey for short) Of SchlockMercenary fame of course.

[ShadowSonic]

790, from Lexx.

Kryten, from Red Dwarf.

It's the laughs that count in life, why not give credit where credit is due?

[NecronLord]

Be fair. 790 was briefly a magnificent warrior von-neumann swarm, too.

[The Yosemite Bear]

Roy batty and R. Daniel are still the best for good or ill. Yeah, despite being a homocidal sociopathic machine Batty just has that great death scene and the king kong "we who made him that way are the real monsters" thing going for him. then yyou have good old R. Daniel, the Asimov uber-robot, that couldc take dcown ancy of the johnny come latelys in such a painfull way that it would look like Durandal raping Shodan seem fair.

[Typhonis 1]

Has anyone mentioned K-9 yet?

[CaptHawkeye]

A best robots thread and not one person has mention KOS MOS yet? Damn.

Aside from the fact that she just looks and acts cool, she is perfectly feasable for any situation you may be in. Hell, she may be overkill.

-She has a built in armory. All of her weapons are built into her frame.
-Her power generation is enormous. She was able to wipe an entire fleet of thousands of Gnosis in one shot. Bear in mind, Gnosis have observed shielding capabilities.
-She can generate a hilbert wave over several AUs.
-She has observed shielding capabilities. (Although they may be limited to certain versions of her armor.)
-She's durable. She was able to withstand quite the beating from T-Elos and Omega for some time and continue operating. Even while being slammed through armored walls, rammed with school bus sized stallagtites, and punched by an Evangelion size fist repeatedly.

[Nyrath]

What would happen if Marvin were to interface with a Culture Drone and spend time telling it about its life philosophy?

The Culture Drone would deduce the dangerous effect of Marvin's speech after listening to three words, it would cut off all communication to Marvin, then it would re-package Marvin as an AI killing weapon to be used against less advanced cultures.

[haard]

Pah. Ways of Earth beats them all!

[haard]

Pah. Ways of Earth beats them all!

Ghetto ed: I realized this may be too obscure. Source is Dark side of the Sun, (T Pratchett.)

Ways is a class 5 (i.e. fully sentient, considered to be 'Human') robot assassin, designed to be lucky

Will supply more data if anyone cares enough to ask for it =)

[VF5SS]

A best robots thread and not one person has mention KOS MOS yet? Damn.

Aside from the fact that she just looks and acts cool, she is perfectly feasable for any situation you may be in. Hell, she may be overkill.

She's also really really hot. Like, smoking. She's also a progressive role model as Shion's lesbian lover.

[Lancer]

Hedonism-bot.

[Crazedwraith]

Pah. Ways of Earth beats them all!

Ghetto ed: I realized this may be too obscure. Source is Dark side of the Sun, (T Pratchett.)

Ways is a class 5 (i.e. fully sentient, considered to be 'Human') robot assassin, designed to be lucky

Will supply more data if anyone cares enough to ask for it =)

Ways Luck function was totally awesome. But I prefer Isaac. Simply for the 'sitting duck' scene with the army of class 3 robots.

[CaptHawkeye]

A best robots thread and not one person has mention KOS MOS yet? Damn.

Aside from the fact that she just looks and acts cool, she is perfectly feasable for any situation you may be in. Hell, she may be overkill.

She's also really really hot. Like, smoking. She's also a progressive role model as Shion's lesbian lover.

I still wonder to this day why Kevin Winnicott would design an elite battle android with boobs. They even bounce...

A lesbian lover for Shion? A possibility I guess. Isn't she busy with chaos though?

[VF5SS]

I still wonder to this day why Kevin Winnicott would design an elite battle android with boobs. They even bounce...

A lesbian lover for Shion? A possibility I guess. Isn't she busy with chaos though?

If I may quote this random blog.

But back to the important bits - Shion and KOS-MOS. Someone on the
Yuricon Mailing List commented that any woman who makes a robot that looks like an attractive woman, complete with sexy outfit and garters, is immediately suspect.

The sexiest part about KOS MOS is her heels. She does battle in sexy heels and it's awesome. Not that I've ever seen the Xenosaga show or played the games, but I know awesome when I see it.