Archives

Filters

Pearls of Weber

A collection of posts by David Weber containing background information for his stories, collected and generously made available Joe Buckley.

Artificial intelligence and nanotechnology

The question of AI in the Honorverse has reared its head once more. Is there AI, and if there is, why hasn't it been seen centerstage?

 

There's not what I think of as fully developed artificial intelligence in the Honorverse. What there is might be thought of as highly developed "expert" software packages. At the same time, there's an enormous amount of what would probably be called AI (although not strictly correctly so, as I understand the term) which is subsumed. As you know, I tend to assume that a lot of the standard science-fiction high-tech goodies are going to be so much a part of the background and infrastructure of the characters in my novels that I let them sort of disappear into the background. It seems to me that someone living in Honor's time would spend as much time thinking about nanotechnology, let's say, as someone in our own time spends thinking about the technology used in bringing us frozen or freeze-dried food products. It's just there, doing its thing very quietly in the background, without attracting a great deal of notice.

The same is true of my approach to dealing with AI. If you go back and look at the battle sequences, you frequently find passages like "no human brain could have kept track of the situation" and like that there. That's because, quite literally, no human brain could keep track of things once the ships really get into the furball. What happens at that point is that the "AI" which is always there, coordinating the ship's targeting, fire distribution, communications links, point defense, etc., steps in and takes over total control within the parameters of the instructions it's received from its human command crew.

Under normal circumstances, the expert software running ships in the Honorverse is designed to accept human input and to be overridden by human command crew at moments of the command crew's choice in order to maximize its responsiveness and to get trained human intuition into the decision-making loop. The less capable the "AI" in use, the more human input is required, which is one reason that Havenite tactical officers usually actually have more to do than Manticoran tactical officers. Their "automatic" systems are less responsive and less capable, which requires more direct human supervision. This is also what's really going on with a lots of the Alliance's "automation" of their warships. The traditional warship design philosophy emphasized human redundancy in the control systems and a sort of "decentralized" approach to the use of computer control. Merchant ships, although they have much less capable electronics, have always used "AI" to a much greater extent than warships have, because they are not designed and intended to take massive damage and remain combat-capable. Warships are designed with that philosophy in mind, and the nature of the tasks which their control systems have to manage is more complex by orders of magnitude. [What the Manticorans have done is to accept an even greater reliance on the "AI" people seem to be missing, with a consequent reduction in redundancy which may end up equating to a more rapid erosion of combat capability in the face of damage. Obviously, they've tried to protect themselves against this by incorporating improvements in automated (computer- controlled) damage control systems, repair remotes, etc., but the fact remains that they have accepted a higher probability of attrition in combat effectiveness in order to get the greatest combat power they can out of a given platform's tonnage and a given, lower number of personnel. The core systems -- electronic warfare, engineering, etc. -- have relatively denser manpower allotments even under the new dispensation because they are the ones critical to the ship's survival, and therefore are assigned a greater degree of redundancy. Manticoran ships are more likely to lose offensive capability faster than they will lose defensive or propulsive capability, if that makes sense.]

It's important to bear in mind when thinking about "traditional" redundancy of design, for example, that although energy mounts and missile tubes have on-mount crews and local fire control, those on-mount crews and fire control systems are provided solely to be used when the weapons have lost their links to the centralized, computer-managed and directed tactical systems. The tactical officer sets up the parameters of his tactical plan, allocates his resources in broad terms in response to the direction of his commanding officer, and then hands over to his computers, which execute his plan. Normally, there are points in the computers' decision trees where they actually have to modify the instructions they were given in light of changes in the tactical situation. What the human tactical officer is actually often doing in the middle of an engagement is looking at the alternate decisions which the computers are offering up for his consideration. As long as the pace of the action is slow enough, this direct human participation in the decision-making loop is practical, and the consensus among the various space navies of Honor's time is that the human element leads to superior performance as long as the time for it is available. Once the pace of the action moves beyond the level at which human operators can keep track and they begin to fall behind the decision-making curve, they normally hand off to the computers. The problem Thunder of God had in Honor of the Queen was that her AI was far inferior to Fearless' to begin with and that her tactical officers didn't really understand when and how to get out of the way. When they finally made the decision to hand over to computer control, effectively completely, their offensive fire was vastly more effective than it had been while half-trained human operators were getting in the way. Where they got into trouble was that they were unable -- because of their inexperience -- to recognize that their idiot Havenite artificial so-called intelligence was repeating its defensive tactics in a predictable cycle, which Fearless' systems recognized and brought to Rafe Cardones' attention, requesting human operator input to decide what to do about it. At which point Rafe blew the crap out of the vastly more powerful ship.

The same situation -- generally -- obtains for point defense. Basically, the missile-defense officer sets up a defensive plan, which the computers then execute. Once again, however, the system is overseen at every stage by its human operators. In an emergency situation, point defense goes to a completely computer-controlled state, as when (again from HotQ) Rafe uses his laser clusters -- under computer control -- to take out the ambush missiles launched by the Masadan LACs. When a missile-defense officer is described as watching his/her displays, with "fingers flying" across his/her console, this should not be construed to indicate that he/she is actually making missile-by-missile decisions for counter-missile launchers or point defense clusters. The missile-defense officer is engaged upon an intense electronic conversation, if you will, with the automated control systems executing the missile-defense plan currently in operation. Refinements, queries directing the computers to consider different analyses of incoming missiles' EW systems and strategies, overrides releasing missiles from shipboard control sooner than the computers would have because trained human instinct and intuition is balancing lower probabilities of kills than the original missile-defense plan specified against the needs to engage greater numbers of targets in the face of heavier fire, etc., is a significant part of the missile-defense equation, but the computers themselves are quite capable of conducting a missile-defense fire plan without human intervention if they have to -- or, again, if the tactical situation has become so complex that the human decision maker's inability to process and evaluate information is overwhelmed. (By the way, you will get to see in Shadow of Saganami an example of just what "point defense" lasers can do at short range when used in the anti-ship role.)

All of these things have happened repeatedly in the books. I simply have not broken it all down detail-by-detail. It's similar to the way that I have the helmsman on one of these ships using a joystick. He/she isn't literally using the joystick to "steer" the ship by eye, using purely human reaction times and hand-eye coordination. The joystick can be used in that fashion if the ship has taken heavy combat which has damaged the computers normally responsible for handling the mechanics of maneuvers. That, in fact, is exactly what Honor does with the first Fearless in OBS. When she takes over the helm, she is flying the ship by hand -- she is, after all, Honor Harrington, so she can do that kind of thing; which is not to say that any mere mortal could do it [g] -- but normally, the "joystick" is actually functioning as a very complex multi-function input device, a sort of direct spiritual descendent of the "Hands on Stick and Throttle" concept being used today. A helmsman has several options for how he is going to use it. For example, he can (and normally does) use it to drop a cursor into his maneuvering plot, then input acceleration settings through the manipulation of various control buttons, at which point the ship's computers takes the ship where he's told it to go at the acceleration settings he's told it to use. Now, in smaller vessels, like pinnaces or Honor's runabout, the joystick is far more frequently used in a fashion very similar to that of present-day aircraft control systems, and LACs sort of fall midway between small craft and starships in this area.

While I am on the subject of AI, I suppose I ought also touch on nanotechnology, although I think I've done this somewhere else, since there are similarities in the way I relegate both of them to "background."

Basically, there is a lot of nanotech in the Honorverse. It's very heavily used in medicine, for example. I've referred to it in assassinations, in Allison's discussion of genetic engineering, in the fact that specifically targeted nanotech is used instead of antibiotics, etc. Now, the nanotech in the Honorverse is much less capable than the nanotech in Prince Roger's universe, which reflects in part certain prejudices of mine -- or, perhaps it would be better to say, reservations -- about things like energy sources and waste heat, and like that there. Although most of the specific appearances of nanotechnology in the books have come in the life-sciences fields, that is mainly because, if you think about it, I haven't spent a lot of time in the Honorverse's heavy industrial sector. There should, however, have been certain clues indicating that nanotechnology is actually used even more extensively in industrial applications than in medical buttons. I've referred to the fact that warships' armor is "grown" in place, I believe, and the differences between Grayson's pre-Alliance technical infrastructure and what the Graysons are capable of as of the latest books has a lot to do with the replacement of pre-nanotech construction techniques and abilities with the fully developed Manticoran version. Let's think about the size of the ships that we're building in a year or so in the Honorverse. Obviously, there will have been enormous advances in "normal" materials handling technology, and they have the advantage of building in microgravity, but they're also using a lot of nanotech to "grow" control runs, seamless bulkheads, the armor already referenced, etc. Warships are also designed around the notion of sheer brute force applications for repair purposes, because they face the probability of combat damage which has to be worked around.

And while I'm on the subject of nanotechnology, remember the "smart paint" used by most of the Honorverse navies. That incorporates quite a bit of nanotech, and, among other things, is used in management of waste heat by Honorverse ships. Whenever an impeller-drive vessel's wedge or sails are up, they are part of a complex energy management system. They are able to "siphon" energy across the interface to "higher" levels of hyper-space, and they can also be used as an energy sump into which waste heat can be dumped. A lot of the waste heat produced by these ships is recaptured and converted into stored energy in the capacitor rings, but there is no way in the universe that they could recapture and use enough of it to keep from cooking their crews under normal circumstances. The "cooking component" heat is what's routinely radiated into and through the wedge and sails when a ship's nodes are hot. When they aren't hot, other arrangements have to be made. Merchant vessels, which aren't armored and do not normally use "smart paint," incorporate radiator arrays into their basic hull design for use when the wedge is down. Warships, which do use "smart paint," effectively reconfigure their "paint" to convert it into a heat-radiating surface.