08-02-2018, 06:05 PM | #1 |
Join Date: May 2010
|
Keeping humans relevant in the shadow of TL10 AI.
One major omission is Ultra-Tech is any sense of the availability of AIs with various skill levels. This represents a huge unanswered question about any ultra-tech setting. If AIs with skill-18 are ubiquitous, there's nothing for humans to do. But Ultra-Tech seems to assume that TL10 societies will still be fairly recognizable, with human obsolescence not becoming a real threat until TL11 or even TL12. But how do you enforce this in-game?
Transhuman Space vaguely gestures at ways of doing this. The first Personnel Files book is explicit that AIs with skill levels above what it calls "ordinary expert" are rare. In practice, many off-the-shelf AIs have as few as 4 points in their main skill. While AI software can theoretically have complexity as high as 10, I'm only aware of one stray reference to such a powerful AI in any of the sourcebooks. Most of the examples of AIs mentioned are NAIs and LAIs that are at best as smart as the average baseline human. The smattering of SAI examples range from SAI-7 to SAI-9, with the latter being implied to be mostly in use in the military or cutting-edge research facilities. Ultra-Tech uses slightly different rules for AIs, but you could do something similar. Make AIs more complicated than a fast non-volitional AI with IQ 10 (a Complexity 8 program, requiring a microframe or fast personal computer to run) at least unusual. Don't be afraid to limit AIs to 4 points in their main skill, and 1 or 2 points in other skills. You can make exceptions to the 4 point rule - maybe 8 points in some Electronics Operation specialty is common because most electronics in the setting "just work" even for untrained users. Thoughts on this approach? What have other people done? At some point, I want to try working out the effects of this approach on the role of drones in spaceship combat, but that's a project for another day. |
08-02-2018, 07:46 PM | #2 |
Join Date: Feb 2016
|
Re: Keeping humans relevant in the shadow of TL10 AI.
I generally avoid using AI but, when I do, I make AI a function of hardware rather than software. An AI can only emerge from a neural net and, since the evolution of each neural net makes it functionally unique, it cannot run on any other neural net, meaning that each AI is bound to their resident neural network (though they can upgrade their neural network). In order to reflect the utility of AI optimization of their resident neural networks, I rule that neural net computers with resident AIs are capable of running an additional program at the neural net's Complexity for every point of IQ possessed by the AI, so even very dumb AI are quite useful to corporations.
Without the ability to circumvent their physical bodies or to create duplicates, AI become much like any other character, though they have certain advantages and disadvantages. Very intelligent AI require very large computers, so they are incapable of traveling with the party. Smaller AI are more portable but, since they are vulnerable to damage and cannot heal, the majority of them are very cautious and are a little dumb. |
08-02-2018, 08:12 PM | #3 |
Join Date: Sep 2004
Location: Orem, Utah, USA
|
Re: Keeping humans relevant in the shadow of TL10 AI.
Truly volitional AI aren't inevitable they're just what some people think might happen with enough processing power. If you assume all AI have the Automaton meta-trait they go from being super smart beings that overshadow the humans to extremely effective tools of the humans.
|
08-02-2018, 08:14 PM | #4 | |
Join Date: Aug 2004
Location: Buffalo, New York
|
Re: Keeping humans relevant in the shadow of TL10 AI.
Hi Michael Thayne,
You raise a valid point - one that comes into play largely because of assumptions being made about complexity values for IQ and so forth. Scientists were always certain that Artificial Intelligence was just around the corner, but we have yet to see it. We can see AI skill systems, but what we generally don't see are free floating AI skill systems that are combined with a whole bunch of other skill data sets. Then came that fateful day when GURPS ROBOTS was published, and ROBOTS could be built as characters with an engineering component to them. The biggest issue to my point of view (one that isn't shared by many, but more on this later) - is that Once the robot was created (as a character) it could GROW as a biological. In other words, the IQ stat could be improved separate from the Complexity of the original hardware. Toss in the fact that Eidetic Memory gave more of a bang for the buck than they should perhaps have done - and the AI characters could suddenly end up with insanely HIGH skills. Now, I mentioned up above that there would be "more on this later" and this is the later part. GURPS ULTRA-TECH for 4e made a profound change in the rules. A computer's IQ can NEVER exceed its initial limits for IQ based on the computer's initial complexity. In other words, if it starts at IQ 10, and that's the highest it can go because of the complexity of the computer itself, then that AI character can never improve its IQ any higher than 10 with character points. Which brings me to the final point. GURPS 4e rules for robotic characters (which would include NAI, SAI, etc ) doesn't really limit how much in the what of character points - can be invested in skills. There is no Hardware/Point cost system for 4e as there was with 3e. Making matters worse if the fact that there are no real rules for how one gains experience within a given skill beyond a given point. One houserule that I use for my own games is that OJT can never raise your skill above a 14. Day to day use of a given skill under ordinary circumstances - are just that, ordinary circumstances. One needs to be exposed to some pretty extreme situations or rare circumstances in order to increase one's breadth of knowledge. Either that, or one has to accept the tutelage of someone else who has amassed that particular piece of knowledge that is uncommon. To become truly expert, would require that you have a grasp of a LOT of the uncommon aspects, which would require in part, that a person either experience the unusual events or read of those who have and collect their tidbits of wisdom to accumulate with your own storehouse of wisdom (knowledge). So how does an AI actively acquire a skill in excess of 14? What is required in real life to qualify as a skill 15 (as opposed to having another 4 character points in a skill game mechanics wise)? In the end? The moment you allow robots and AIs to be purchasable with character points - they become a new "race" that is in direct competition with Humanity. Robots and AIs are no different than Elves or Pixies or what have you. Should a GM be trying to make Humans relevant in a fantasy world where Elves are better and more predominant and the top of the apex predators? Not a philosophical question I can answer. Ask 10 philosophers one question and they will give you probably 20 different answers ( despite there being only 10 people answering!). My suggestion is this - take a page from the original rules from ROBOTS. Each "Skill" constitutes one "program" in the computer. The more skills you want your AI to be able to learn, the more programs that have to be running to be able to access all skills at all times. The more points that go into the skill, the higher the complexity of the software, which will soon run into the limitation that any given AI computer can only run 2 programs of the max complexity, 10 programs of Complexity -1 plus 1 program of the same complexity etc. This is an excerpt from my thread on how to convert GURPS ROBOTS for use with GURPS 4e (well, actually, it is about making the computer rules for 4e fit in with the rest of the GURPS ROBOT design rules) Quote:
So, if an AI is to have a skill 14 with an IQ 10 computer brain, that becomes Stat+4. For an easy skill, that costs 12 points. For an Average, that's 16, for a Hard it is 20, and for a Very Hard skill - 24 points. That works out to a complexity 5 program for easy, complexity 6 program for an average, a complexity 7 program for Hard, and a complexity 8 program for very hard. I also use the rule that you can't copy a neural net program and install it in another neural net brain simply because they aren't the same computer and won't evolve the same way. When you buy a brand new AI with a Neural Net and you give it 10 chips for storing its skills in - and the computer is say, a complexity 7 computer, it will only be able to have 10 skills, of which none of them can be higher than a skill with 20 points in it. All other skills have to be at least a complexity 6 level program. Again, this is all up to you. I would suggest you read the thread on how to convert material for use with 4e from GURPS CLASSIC ROBOTS for more of what I'm talking about. Might be worth your while. |
|
08-02-2018, 08:54 PM | #5 | |
Untagged
Join Date: Oct 2004
Location: Forest Grove, Beaverton, Oregon
|
Re: Keeping humans relevant in the shadow of TL10 AI.
Quote:
Using THS terms, you don't need to only use NAI, or not at all sapient programs or fully human SAI or sapient programs. There's a huge range in between. As long as you keep setting humans competitive with setting A.I. or at least far more populous things could be stable at least for a while.
__________________
Beware, poor communication skills. No offense intended. If offended, it just means that I failed my writing skill check. |
|
08-02-2018, 09:06 PM | #6 | ||
Join Date: May 2010
|
Re: Keeping humans relevant in the shadow of TL10 AI.
Quote:
Quote:
|
||
08-02-2018, 09:10 PM | #7 | ||
Join Date: May 2010
|
Re: Keeping humans relevant in the shadow of TL10 AI.
Quote:
Quote:
|
||
08-02-2018, 09:19 PM | #8 |
Join Date: May 2010
|
Re: Keeping humans relevant in the shadow of TL10 AI.
One thing about Transhuman Space is that yes, on the one hand you do have 18,000 ton warships that only use six human crew, infantry units where every human soldier is in charge of multiple serious combat robots, and "supervisor of non-sapient robots" as a type of menial labor. On the other hand, it has many worked examples of adventuring parties (in the Personnel Files line) that are conspicuously not like that—the humans are mostly limited to not-very-smart wearable AIs and maybe an odd robot in a supporting role.
|
08-03-2018, 08:49 AM | #9 |
Hero of Democracy
Join Date: Mar 2012
Location: far from the ocean
|
Re: Keeping humans relevant in the shadow of TL10 AI.
Here are a few key things to remember about TL10 AI and computing:
Of course, when I run a TL10 setting, I carefully decide how AI works in this setting and alter the allowed technology to fit this vision. Some things I've used or considered include:
__________________
Be helpful, not pedantic Worlds Beyond Earth -- my blog Check out the PbP forum! If you don't see a game you'd like, ask me about making one! |
08-03-2018, 08:55 AM | #10 |
Join Date: May 2010
|
Re: Keeping humans relevant in the shadow of TL10 AI.
As far as spaceship combat goes, if drone AI only has skill 11 while it's very possible to train humans up to skill 14, that's a non-trivial advantage, but it doesn't really change the fundamental problem with the combat side of Spaceships in that missiles are still ridiculously under-powered. If you're firing missiles at a SM+8 frigate, base skill 11 + 2 (from sAcc) + 8 (from target SM) + 4 (from proximity detonation) is effective sill 25, which virtually guarantees 10 hits.
Alternatively, it lets you soak up huge penalties from a high-speed attack run, which in turn increases damage. In tactical combat, it's quite easy to arrange missile attacks with 10 mps relative velocity. That's only a -6 penalty, so your effective skill is 19. You'll still score 8-9 hits most of the time, while making a 16cm missile inflict 840 points of damage! So spaceship combat is broken, what else is new? It was probably over-optimistic to think limiting drone AI would fix this. You could go with significantly more limited AI, but this ends up being fairly implausible. One thing Spaceships never really reckons with is that while worked examples of drones just remove control stations, Ultra-Tech microframes weigh less than a human pilot and cost less than 2% of what the cheapest drones cost, so there's no reason for drones not to have microframes installed. At TL10, this means a drone can have that IQ 10 fast non-volitional AI running. Even with only 1 point in Artillery (Guided Missile), that's still skill 9, which is adequate (if not ideal) for high-speed missile attack tactics. |
Thread Tools | |
Display Modes | |
|
|