A robot is in a weird classification somewhere between a vehicle and a person. This makes them tricky to handle in a lot of RPG's, and Gurps is no exception. One of the hardest challenges is that both characters and vehicles ST and HP, but characters get DX, while vehicles have handling. In this post, we're going to explore how to make machines into characters. We essentially do this by separating the AI pilot from the vehicle they are driving.
Everyone is a PilotAs I said, the key to making all this happen is to treat the robot bodies as vehicles. This separates characters from equipment, eliminating much of that tricky grey area. In Robots as Spaceships, vehicles are driven, not inhabited. It doesn't matter if they are steering it using a wheel, piloting it remotely from the other side of the world, or directing it electronically AI from their onboard server. Everything is done as though the character is a driver or pilot.
This has a couple of important effects, especially for AI's whose hardware is stored on the vehicle.
First, the DX of an AI is determined by what software its running, not the speed of its servos. Yes, some vehicles are faster and more agile than others, but these can be represented with handling, stability, move speeds, and tool quality modifiers. A skilled wheel-man doesn't loose his reflexes because he's sitting in the seat of a bulldozer rather than a motorcycle. He'll be able to get more out of the lumbering vehicle than the nerdy hacker precisely because he's got better reflexes. Similarly, a dedicated racing program will get better physical maneuverability than an AI built for analyzing astronomy.
Second, the HT of a vehicle is separate from the HT of the box containing the AI. An AI carried in a vehicle that is shot to pieces may not even get hit. It should also be noted the ST of an AI is not connected to the body its currently in. In fact, AI's won't have ST scores at all! This will have implications later on, but we're not going to explore them right now. For the moment, the ST score of an AI is "none" and neither costs nor gives back points.
What Should the ST of a Robot Be?So what's the ST of the vehicles? When we looked at an SM+0 gunbot, we discovered that 30 is high, and 20 is low. We want to stay on the size/range chart, and there is unfortunately no in-between option there. And come to think of it, there shouldn't be a single correct ST for every robot of a given weight. Of course, there isn't a single ST that all robots of a single weight should have: higher tech and more expensive robots will have better power to weight ratios.
So we're going to step down the expected ST of a robot by one step, from 30 to 20. However, the following lenses can modify ST:
Fine ST robot (+25% ST, +100% to chassis cost)
Very Fine ST robot (+50% ST, +300% to chassis cost)
For now, we will not add a scheme to increase ST with Tech Level, but doing so is not unreasonable at all.
A Mechanical MindThe easiest statistic of an AI to determine is IQ. Indeed, Ultra-tech AI's are basically defined by their IQ and their sapience level (Dedicated, Non-volitional, or Volitional). We may later find reasons to lower IQ from the basic complexity stats, but IQ will default to complexity.
By default, Will and Per are based off of IQ, and we're going to keep things this way. Some robots are unusually resistant to negotiation and fear, but these are usually the result of a trait (Indomitable or Unfazeable), not of superior will. Perception is also likely to scale with IQ, as both increase with processing power. Perception will be modified by sensors, of course. We will be allowing Ultra-tech sensors to have the same bonuses for AI's as for humans.
There are somethings an AI can be programmed for that are rare in humans, like the ability to track multiple "views" at once, but we will cover than in a later post: its nice to have, not essential.
The DX of a MindJust how agile should a robot mind be? Do they feature in inhuman reflexes of a targeting system? The lumbering fumbling of a bulldozer? This is ultimately a setting call, and a GM should pick a DX scheme that works for him.
Surprisingly, most of the time, the DX of a machine should be roughly human. Most systems that enhance the targeting of a robot can be used by a human as well, or they're a result of instant decision making. Enhanced Time sense takes care of most of these cases. In most fiction, robot's aim roughly equivalent to their human counterparts. Their dodge is often low, but this is often a feature of their chassis, not their programming.
Better reflexes is generally going to require more processing power. While simplicity can occasionally make for faster action, DX is more about complete coordination. We're not sure what the best scheme for determining DX is, but there are a few different schemes we can try:
Matching IQ: The simplest approach is to simply give the robot DX equal to its IQ.
Shadowing IQ: DX equals IQ -2. As an option, swap the two attributes, for fast killer bots that are really not that bright. Per should be based off the higher of the two values.
DX-IQ continuum: DX starts out at 10. IQ points may be swapped for DX points. This method is not recommended for IQ's that start below 10.
Each method has its upsides and down sides, and some of them suggest other schemes. All of them should work, but only one should be used in a given game, unless you use matching or shadowing for IQ below 10 and the DX-IQ continuum for IQ above it. For our default system, we'll start by shadowing IQ, but we'll need to test things to determine which method works best.
Buying SkillsThe hardest part of building a robot is how to price skills, mostly because ultra-tech doesn't give any guidelines. We are going to allow software tools (page 25) to represent an AI knowing a skill program. Basic programs allow them to know a skill at the +0 level. Good and Fine quality programs raise their skill by +1 and +2 respectively. These programs are NOT the Good and Fine quality programs humans use, but they are similar enough that Artificial Intelligence cannot claim a bonus from using that software.
I should be noted this purchase system does not lend itself to Heroic Robots. This is because such entities should be exceptional, and discovered or developed rather than purchased from a rack. Heroic Robots are better built with points than with equipment, at least as far as skills are concerned. The vehicle part of this system will allow heroic AI to do awesome things, but if you want an AI standing out as exceptional, the last thing you should do is limit it with factory specifications.
Healthy as a Robot?
There are two different HT scores on a robot built this way. The first is the HT of the vehicle. By spaceships, a spaceship SM+4 or smaller will have HT 12, or HT 11 if its totally automated at TL 7-9. We will probably talk about this more in a different post.
The HT of the Artificial Intelligence is probably the generic 12 that Gurps gives most objects, but its worth considering if that's what we really want. We should also consider what an AI uses its HT for. In most cases, AI's will use HT to resist Hacking, Viruses, and other electronic attacks. Will has often been used in Gurps to represent a resistance to Hacking, under the reasoning that its a mind-influencing effect and thus will is good at resisting it. Of course, that's not how hacking works. An AI's resistance to digital manipulation is a function of how securely its built, not how good it is at resisting social influence or how dedicated it is to a task.
Bodies and VehiclesWe've taken some time today to talk about Robots as characters, and what attributes we want on our vehicles. This is not the one true way to build robots as characters, but its important to be consistent. We now know that this series will treat AI's and the shells they inhabit as separate entities. This will make the rest of our process much simpler.
I also hope I've given an option for GM's struggling to run robots and biological characters in the same game. Mind-hopping can be difficult to handle in a game, and the separation between mind control and piloting a vehicle can be very subtle. I hope someone finds this approach greatly simplifies their game.