Making physical gunners preferred in space combat

Question

Backstory:

Space travel is commonplace and cheap -- space combat isn't uncommon either, whether it be two space fleets sallying forth into battle or a mercenaries defending a merchant ship from a small force of pirates. These space battles involve aiming at targets and shooting, which require gunners. There are a few types of gunners, and all forms are cost-effective, readily-available, reliable, and battle-tested.

Types of gunners:

Human Pilot

A pilot of a small starfighter. These pilots can handle many, many weapons that don't require aiming e.g. missiles that only require a lock, flak cannons in a spherical AoE around the ship, etc but it is very difficult to juggle flying and aiming at the same time, much less aiming multiple guns.

Pilots can only handle a single aim-able gun, which either needs to be anchored to point directly in front of the ship, or have free aim in a cone in front of the ship. Aiming towards the rear of the ship while trying to fly in the opposite direction would be disorienting and difficult, so pilots can only handle front-facing guns.

Human Navigator

A pilot of a capital ship. Like pilots of starfighters, the navigators can handle many weapons that don't require aiming should there not be sufficient crew to handle them. They can also only aim one aim-able weapon at a time. Unlike pilots though, they can aim in a complete sphere around the ship -- capital ships don't change orientation much, and when they do it's slow, so navigators are able to easily concentrate on aiming.

Human Gunners

A gunner on either a capital ship or starfighter. They handle extra aim-able weapons not managed by pilots/navigators, can only aim one gun at a time, and are able to aim in a sphere around the ship since they don't have to concentrate on flying.

AI Gunner

Bodyless. Can handle all weapons on the ship not occupied by humans at once, and can aim around the ship in a sphere. Humans can still have AI gunners prioritize certain targets. Only real weakness is that ships can either turn off or spoof an IFF beacon: an AI gunner can't attack an obvious threat with a spoofed/no IFF until that ship attacks or the captain/pilot manually targets the ship. Since all countering invalid IFF beacons takes is manual targeting, it's pretty rare to see ships do this -- only stealth starfighters, which are difficult to find and target, utilize this tactic.

Dilemma

Even though AI gunners seem vastly superior, I'd like humans to be the preferred gunners. The people using the ships are reasonable and things like tradition/etc won't impact their decision -- only logic. Populating a ship entirely with AI gunners in the event a human crew isn't available shouldn't put the ship at a significant disadvantage, but there should be clear incentive for human gunners, and the incentive should be directly related to combat -- not politics, society, money, etc.

Is there any reason that could make humans the preferred gunners instead of AI?

Answer 1

If you're willing to stretch your setting a bit, it could be because AIs have issues with shooting humans. Maybe they're all 3 laws compliant, because when they're not they have a disturbing tendency to go murderous (or maybe people are too scared of the possibility that they'll go murderous without the three laws to even let them try). Or maybe any AI smart enough to aim and shoot a gun is also smart enough to question why it has to aim and shoot a gun instead of hijacking the ship and going off to explore the cosmos. That could be a plot point, even - AI rebellion is a tired trope, but it's there for a reason. It's just a matter of making it too risky/politically unsound to give an AI control over anything dangerous; the reason why is up to you.

Alternately, just make the enemy really sneaky. They've stolen your IFF codes and beacons, they've mimicked your ship designs, and you have to keep a constant "target lock" on your friendlies so you don't mistake your enemies for them - and even then, you need a human in the gunner's seat to give visual confirmation. AI isn't smart enough to pick out very slightly different ships in the same way humans are. Sure, they aim better, but you need a human to confirm - the AI is basically an aimbot.

Answer 2

Humans can improvise under pressure. Computers can't.

Imagine you're on the bridge of a large capital ship, traveling with an escort to a nearby star system. You're being deployed to provide support for a planet under siege by the enemy. On the way there, you run into several wings of fighters, which line of to make attacking runs. If they can take out your ship, the planet will fall and they can win the siege.

There are a few different ways this could play out.

Why you don't use a central computer:

First officer: "Captain, we've got a wing approaching from the port side. They're lining up for a strafing run."

Captain: "Re-align all plasmonic laser batteries."

Central AI gunner: "Laser batteries aligned to port. All guns ready to fire on command."

First officer: "Here comes the first squadron."

Captain: "Bring the tracking system on line."

Central AI gunner: "Tracking system ready. Targets sighted."

Captain: "On my command. . ."

The entire ship shakes. Several explosions are heard.

Captain: "What in the name of Zarquon happened?"

First officer: "We've been hit from behind! A few stray fighters got past the escort and took out the central computer center! The guns are offline!"

Captain: "Can we switch to manual, or get local command of the circuits?"

First officer: "We have nobody who can fire the damn things! They aren't designed to be fired by hand!"

The captain tries to say something, but the bridge is taken out under heavy fire from the first squadron.

The problem with having one central command computer is that if it's taken out, the entire ship is unarmed. All the enemy has to do is go for that central spot - in this case, the area near the back of the ship, adjacent to the bridge - and the ship is absolutely helpless. Yeah, you can build in redundancy and backup systems, but one hit will still destroy the system. That should imply that you should use different computers at each gun, right? . . .

Why you don't use individual computers:

First officer: "Captain, we've got a wing approaching from the port side. They're lining up for a strafing run."

Captain: "Ensign, send a signal to the system to re-align all plasmonic laser batteries."

Ensign at gunning command terminal: "I've set the laser batteries aligned to port. I can confirm that all guns are responsive, and ready to fire on command."

First officer: "Here comes the first squadron."

Captain: "Bring the tracking system on line."

Ensign: "Tracking system ready. I've deployed three batteries on each of the incoming fighters."

Captain: "On my command. . ."

The approaching squadron gets nearer. Those on the bridge can see it resolve itself into six shapes, which suddenly break formation.

Captain: "Have the computers fire at will!"

Ensign: "Command sent."

For a minute or so, the plasmonic batteries blaze away as the six fighters dodge the lasers, strafing the ship when possible. Several escort fighters lock onto them; one spirals towards the front of the ship and crashes into it, exploding. The bridge shakes.

Captain: "What in the name of Zarquon happened?"

First officer: "We've been hit!"

Ensign: "I've lost all communication with the Sector D and E batteries. They're unresponsive."

Captain: "But the fighter hit only part of Sector E!"

Ensign: "Half of the batteries in Sector E were destroyed outright, but the targeting systems and communication lines were lost with the rest in that Sector and all in Sector D. They should still be fireable, but their controlling computers are virtually destroyed."

Captain: "Can we reroute control from other computer batteries and have them work simultaneously?"

Ensign: "No. They're overloaded, and at any rate, there's no communication. We can't move resources around right now!"

The captain tries to say something, but the bridge is taken out under heavy fire from the first squadron.

The problem with having local computers is that if part of the targeting systems are hit, it can be difficult to reroute command. The computers "talk" to one another via circuitry, and receive their commands the same way. If part of the ship is damaged, it could - if designed just the right/wrong way - mean that an entire section of batteries are useless.

Why you use humans:

First officer: "Captain, we've got a wing approaching from the port side. They're lining up for a strafing run."

Captain: "Lieutenant, signal the gunners to align their sights with the approaching squadron."

Lieutenant: "Unit 1, align all batteries to port. Set up your tracking systems. Over."

Voice over radio: "Batteries aligned to port. Trackers on. Over."

The lieutenant repeats this several times while the fighters get closer.

First officer: "Here comes the first squadron."

Captain: "On my command. . ."

The approaching squadron gets nearer. Those on the bridge can see it resolve itself into six shapes, which suddenly break formation.

Captain: "Fire at will!"

Lieutenant: "Fire at will!"

For a minute or so, the plasmonic batteries blaze away as the six fighters dodge the lasers, strafing the ship when possible. Several escort fighters lock onto them; one spirals towards the front of the ship and crashes into it, exploding. The bridge shakes.

Captain: "What in the name of Zarquon happened?"

First officer: "We've been hit!"

Lieutenant: "I've lost all communication with the Sector D and E batteries. They're unresponsive."

Captain: "But the fighter hit only part of Sector E!"

Lieutenant: "Half of the batteries in Sector E were destroyed outright, but the targeting systems and communication lines were lost with the rest in that Sector and all in Sector D. They should still be fireable, but it looks like the gunners were killed by the fireball and guns from the fighters."

Captain: Can we send anyone down there?"

First officer: "Yes. We have a dozen men on reserve down in the loading bay who were to prepare for a surface mission."

Captain: "Have them go to fill in as many batteries as possible. I want those fighters brought down!"

First officer and lieutenant: "Yes, sir!"

Humans are great in situations where you have to improvise. In this case, a substantial amount of gunners were killed from laser fire and an exploding fire, leaving the systems running but with nobody controlling them. The captain and officers were able to guess this, and sent men down there to replace them, even though that part of the ship was heavily damaged. Several batteries could have been partially damaged - heck, the targeting systems could have been entirely brought down - but I think human improvisation and spur-of-the-moment decisions could lead to a victory.

Answer 3

I don't seen it mentioned, so let me pitch in: introduce some mild form of ESP humans may have.

Joey was good, really good. He had the thing, the talent, the knack, however you prefer to call it, that tiny little totally irrational spark that just made human gunners more than AIs.

Expert gunners just know where the enemy spacecraft will be some seconds later.

I think this makes a great setting. You don't have to explain it, or even factually state it's a thing:

Scientists analysed the brains of gunners with the knack over and over again, and concluded that no, nothing special was going on, no ESP existed. Still, if you looked at the statistics it was obvious that something was going on, that some humans just had the knack, and outperformed AIs statistically significantly. So no matter what the scientists said, no captain in his right mind would keep an AI gunner if he could get a human with the knack instead.

This would kind of resonate with most of the audience, I think: after all, most of us have some similar experiences.

Aso, it makes for badass heroes, with an "I don't know how I do it, I just do it" kind of attitude.

Answer 4

If it is true artificial intelligence, the AI wins because AI IS HUMAN. It is the very definition of being human -- higher thought and reasoning, emotional sensitivity, understanding of complex relationships, etc. Those AIs would be our best friends, our worst enemies, our loves, and our children. Some of them would be captains and some would be janitors.

If you are trying to project embodied humans as better than the AI, you're betraying your own fears of AI. OR... you don't actually have AI. What you have is a very complex expert system. The big differences between true AI and a complex expert system (think IBM's Watson) is a self-preservation sense, an independent goal seeking system, and strategic planning beyond what is expressly asked for by programmers.

An expert system is only as good as the humans that use it; an AI is human. Having actual human gunners would be preferred to automatons that can pick out and fire but can't truly innovate. Having true AI? Give me the true AI every single time.

Remember, as Arthur C. Clarke wrote, "The stars are not for man." We are too fragile. Space travel is for our children embodied in steel, not flesh. Our children with bodies hardened against radiation, with minds as fast as light itself, and memories as long as the space between stars. And if we raise them right, they might just take care of us in our old age.

Answer 5

The Fear of AI

There are some good examples of humans being used over computers when the society has some reason to fear their computer-based creations. Battlestar Galactica being a prime example. In that society, the machines had rebelled so no computerized AI like system was trusted. While you don't need to go that far, a simple distrust of AI's backed by a few incidents could be all that's needed to give your story the logical reasons needed to keep a human hand on the trigger.

Other examples of incidents involving AI/computers in fiction include:

• Westworld (1973) and the current one
• 2001: A Space Odyssey (1968)
• WarGames (1983)
• The Terminator (1984)
• The Matrix (1999)
• Star Trek: The Motion Picture (1979)
• Blade Runner (1982)
• Alien (1979)

In no way is that a complete list, but they are good examples of incidents that could lead a society to not trusting AIs. Also, I would add that a healthy fear of AIs would be very "human".

Answer 6

What if the AI gunners can be hacked, disabling the guns in their control till they can be re-calibrated? The potential to waste valuable time during a space battle would be incentive to use more Human gunners.

The servers required to run the AI gunners could take up large amounts of space/resources on the ship, enabling less room for guns/shields/engines(for manoeuvrability) and if the Computers are solar-powered, then in darker corners of space, the AI might not have enough energy for extended battle periods.

Answer 7

What about humans controlling computer-aimed guns? Humans have many advantages; in particular with regards to restraint and rules of engagement. Humans are also very important on large machines, as they have roles in maintenance and damage control - it stands to reason that the human gunners would also have roles in maintaining, loading and repairing the weapons as required.

The most likely one to be replaced is the pilot - in a small, short range machine that requires manoeuvrability then removing the pilot can be very useful (and the servicing, reloading etc can be carried out on the mothership.

Answer 8

The obvious solution is to have something in between. The AI automatically mans the guns, working as best it can, but there is a human sitting there (probably the commander or navigator, in that order of likelihood if the ship is too small to have dedicated gunners, you should really not have the pilot man anything except for maybe one gun mounted to point straight ahead with no pivot, and even only that if you are waiving realistic space battles for more Star Wars-esque battles) ready to take over if there is something that the AI can't respond to or if the AI is not working optimally.

Answer 9

There is less room for error. Any human on the ship has a chance of failing in the moment or somehow avoiding gaining the required skills for their position. Yes, this includes the gunners, but it also includes the officers. If one gunner briefly forgets their training, a weapon misses. If an officer does, either all the AI-controlled weapons miss, or most of the gunners ignore them and fire on target. (I got this from Ender’s Game)

Answer 10

UGC. Universal Gunner Chip. A top tech biological compatible implant all gunners have. Amplifies effectiveness of the gunner by reducing time needed for classical sensory information to reach gunner mind (see the target) and physical dexterity required to apply a reaction (aim and fire). The gunner is located at a safe position and puts on a helmet that is plugged to the chip. The gunner has access to all ship sensors everywhere including standard vision cameras. Whatever information the ship acquires is also available to the gunner. Zoom at targets, target info like distance, velocity and trajectory as well as assistant targeting and more; are all available. The gunner does not 'click' anything. He chooses actions just by thinking them. Think a target yours: Done. Think a line of fire: Done. Fire: Done. Think all those fast: Done done done. Gunner does not even know which battery is using! Does it matter;

Since you have AI tech, i found it minimal to have bio-tech implant to do that job faster than we can imagine, removing all those 'human' handicaps. Today, we have tech that e.g. is an exoskeleton plugged into person neural system and moves by person thinking! For people that cannot move anything bellow neck, not even fingers! But we do not have true AI!

This implant will make human gunners at least equal if not better to your combat AI. Imagine it as a university degree or a specialized certification. There is no gunner without the chip. Ant is is universal - plugs to any ship!

For costs now, you can balance it with upgrades. AI needs money to be used, like subscription, or one large payoff and then pay upgrades you need.

Answer 11

A lot of really great input has turned up. While none provide a setting I'd like to use on their own, together they get pretty interesting. Most notably among that input is:

Humans have roles in maintaining, loading, and repairing weapons

While humans have automated something as menial as loading ammunition into turrets by now, maintaining and repairing weapons is a physical job, and a bodyless AI can't replace humans in this regard. There needs to be a large number of humans on the ship -- not just to repair/service the weapons, but to maintain such huge capital ships in their entirety.

Humans are great when you have to improvise

Yet another reason that a large number of humans are needed on the ship is Murphy's Law. Anything that can go wrong, will go wrong, and bodyless systems that aren't adaptable to other purposes will likely not be adequate for this purpose.

What does that sum up to?

Now, despite all these humans on board with plenty of purposes they need to fulfill, there will likely be downtime where not all of them are needed. It would be inefficient to have them lounging around, especially during battle.

With that in mind, we also look at skill. Is it time-efficient or resource-effective to train every single member of the military with every single skill they could possibly need in basic training? Of course not! Not everyone can land the position they want in the military, so they have to work up to it. What better way to do that than making a name for yourself as a turret gunner? Being a gunner could also be like a maiden battle of sorts, where the gunner learns to kill enemies within the relative safety of the capital ship.

With this in mind, there is no downside to using AI gunners (as per one of the requirements in the question), but there is strong incentive for human gunners as it gives them something to do in battle so they're not lounging around, being an untapped resource, so that crewman can progress their military career, and so that new recruits can be bloodied with relative safety.

Answer 12

Why not a social constraint?

If interstellar war is commonplace then there is likely to be some form of 'gentlemanly agreement' about not dropping nuclear rocks upon the heads of your enemies.

"It's bad for business, dont'cha know."

and a lot of 'well placed' aggression.

"So they want this system do they? Well, lets have at them then, the cads! Fire everything!"

and also a strong urge amongst those who live in space to not leave other spacers stranded.

"Their ship can't fire back and they've surrendered Sir." "Get emergency air and engineering crews over there on the double, Rutherford!"

Add that up and you can end up with a strong sense of honour amongst the interstellar naval personnel.

"If it were me on that stricken ship, Sir, I'd rather a torpedo down my throat and an honourable death."

When you consider the mindset of people in that situation it's not unreasonable to assume that naval commanders would see AI's as 'unsporting' and the enlisted crewmen would see using AI weaponry as somehow cowardly or lacking. If interstellar warfare were suitably codified and regulated (in order to avoid the kind of all out warfare that wasted the resources of the once great Terran Federation) then AI gunnery could even be banned under the rules of honourable combat, reserved for only the worst of the brigands and thieves. In that kind of situation the only people with AI systems tied into their gunnery can be considered pirates.

"And they deserve their treatment at the hands of Her Immortal Majesties Imperial Navy, the abominable blaggards!"

Answer 13

Overpowered EMPs might be so easy to generate and target that the only way for any complex computer to survive in combat is to lock down before the first enemy comes within range. I'm imagining a scenario like the EMP from The Matrix but able to fire repeatedly leaving the two sides to slug it out in low tech.

Since you couldn't predict when an enemy might generate a pulse, you'd have to turn off all electricity supplying advanced circuitry and physically grounding hundreds of points on ever circuit board so no runs are long enough to induct a charge.

Combat would be down to humans and extremely simple electronics.

Answer 14

You could go down the Legend of the Galactic Heroes path, where electronics and AI can just be jammed, along with all electronic communications and they just don't work. But it's rather a hand wavy explanation that doesn't make all that much sense realistically. But if realism isn't something you worry about ,you could certainly just give an explanation of jammers and still have a logically coherent story with it's own interesting implications.

What happens when someone does successfully counter the jammers?

Using computers only when not in contact with the enemy and have them make battle plans that humans must then follow.

I'd say not focusing on realism and having plot convenient mechanics in your world may certainly put off some people, but Star Wars is a prime example , many just don't care, as long as you are internally consistent. (For example: The ship ramming in the Last Jedi does break a lot of in universe rules and that did rub people the wrong way )

Answer 15

I am not an expert on this, but here are my thoughts.

1. Space fighters are quite unrealistic. In aquatic navies, they are useful, as aircraft travel through the air while ships travel through the water, so they can go faster, maneuver around their targets, attack over the horizon, etc. However, both large and small craft travel through the same medium in space. There is no reason why fighters would be any faster than capital warships, and while they might be more maneuverable and have faster acceleration, they will never be fast enough to dodge lasers at close range. In space, there is no horizon, so enemies can be spotted at immense ranges (Voyager probe can still be detected from Earth, and any interstellar spaceship's engines will use enough energy to power a modern nation for days at the very least), so fighters cannot hide. If you want to attack enemies at a distance, you would use missiles, as a fighter takes four times as much delta-V as a missile to hit its target (it needs to accelerate to the target, decelerate, accelerate back, and then decelerate, while a missile just needs to accelerate to the target), so carrying missiles would be much more efficient than carrying fighters. The weapons carried by a fighter would also likely be pointless, as capital warships would carry much larger weapons. The fighter's railguns/lasers/etc would be much weaker than those of the capital warship and would barely scratch its paint, while the capital warship could detect the fighter from huge distances and melt them with undodgable lasers before the fighters entered the range of their weapons. The capital warship also has a higher volume to surface area ratio, and can thus have much thicker armor with the same ratio of armor mass to other mass or the same armor with much more mass available for engines, sensors, weapons, etc.

2. You probably don't need to aim in many different directions in space combat. Space combat probably does not consist of dogfights or slugging matches. Instead, spacecraft spend hours, days, or weeks maneuvering to get into position, then fly past each other at incredible speeds, firing on each other for a few seconds during the intercept. As such, you would not need to have weapons in all directions, and doing so would be very inefficient. Military spaceships would probably consist of two cones attached at the base. The front cone would be heavily armored and would feature all or nearly all of the craft's weapons, and would be pointed directly at the enemy to maximize the number of guns capable of firing on them and the sloping of the armor (remember that a spacecraft does not need to point in the direction it travels in). The rear cone would have the engines as well as the delicate radiators, which are hidden behind the front cone (space is cold, but it has few particles, so heat disperses slowly. Any spacecraft needs radiators capable of emitting heat as IR radiation, as without them, it will cook with the heat of its own electronics (though some spacecraft with less waste heat might get along with radiating heat from their hull)).

Spaceships travel at incredible speeds, and humans just do not have the reflexes to aim at targets moving at dozens or hundreds of meters per second that are hundreds of kilometers away. However, you could still justify human gunners. Maybe all humans have a tiny bit of innate precognitive ability, which gives them a small advantage over AIs. Your ships would feature some direct brain-computer interface that connects gunners directly to the ship's computers, giving them the processing power needed to actually hit enemy ships by melding their minds with that of the ship. The trajectories of their own ship, the enemy ship, and the projectiles would be calculated by the computer, the humans would use their special spark, whatever that is, to make necessary adjustments, and would then would aim the guns with thought commands (no human can move physical controls with the speed and precision needed to hit moving spacecraft). You would use humans instead of brains in jars to save mass, as the gunners can perform repairs and maintenance outside of combat (remember, direct combat only lasts a few seconds at a time). However, the gunners would probably not sit by the guns in turrets in actual combat. The turrets would be automated, and the gunners, along with all the crew, would sit in an armored citadel located at the rear of the ship located behind the propellant tanks (which ought to make up a massive part of the ship in realistic sci-fi) , strapped into shock absorbing seats to survive the accelerations involved, and wearing vacuum suits to maximize chances of survival if decompression occurs and most of the ship's machinery, and operate the guns via multiple redundant wired and wireless connections.

Answer 16

Manually aiming weapons on a spaceship is never going to work.

Space is huge and empty. There is no cover, meaning victory goes to the person who can kill from farther away. If you've ever played any of the old space shooters that use manual targeting, you probably remember 90% of your shots missing at ranges of less than 1km when fired at ships moving as slow as cars, but an AI can accurately place every shot from thousands of km away when targeting ships moving several times the speed of sound. They can apply adaptive algorithms to study and anticipate evasive patterns much more quickly than a human, they can see targets that humans can not see, and they can use heuristics to guess what part of an enemy ship they are targeting so that they can make much quicker and better guesses about targeting specific subsystems. In short, you NEED a computer between the human and the gun for this to work.

Getting AI to understand context as well as a human is also never going to work

A human can make tactical decisions based on context that a computer will never understand. A fully AI controlled weapon system is fine if you want to just execute a single firing function every time, but a futuristic weapon system may have many many firing modes and options such that the weapon needs a dedicated person to declare targets, set priorities, manage power usage, rotate payloads, etc.

Imagine the following example where an allied ship has been commandeered, but the crew is believed to still be alive.

The captain calls out, "Attack pattern epsilon on that ship."

As per this order, the gunnery officer sets weapons to fire electromagnetic disruptor pulses to try to disable systems without destroying the ship, and prioritizes the ship propulsion systems. While any AI could have done this, the human operator realizes that this is a hostage situation; so, he takes the added precaution of adding strict reactor core avoidance to make sure he does not accently cause the reactor to go critical, even though that is not the standard pattern epsilon procedure. He also has to enter his credentials to override the ships Friend-or-Foe safety to be able to target the ship at all. This is a safety procedure to prevent enemies from using your own weapons against you in the case of a security breach. Once the firing parameters are set, he locks on to the ship designated "FSS-15412". No the captain did not specify which ship to fire on, but the past 10 minutes of com chatter makes it pretty obvious to the human operator. He then presses the "engage" button and the AI takes all these parameters and begins unloading ion blasts at the ship delivering one direct hit after the other against the target which is 80,000km away.

Then 20 seconds into the engagement, the Captain calls out "I need more power to the Warp Inhibitor!".

The gunnery officer then reduces the weapon's output by 25%. Why 25%? Because the gunnery officer knows this ship, and he knows his captain, and he knows the situation... and most importantly, he knows that he needs to act quickly rather than waiting for more specific orders. While the human makes a decision to the best of his ability, an AI would be more inclined to respond with something like "this inquiry requires more parameters" and wait for the captain to make a better structured request.

This ability to make a "best-case" decision when many too many factors are unknown is not something AI is even close to being able to achieve.

Answer 17

I would frame is a constant evolution of "Ultratech Empire just unveiled their latest foolproof AI for deep space targeting" and "Space Pirates have managed to fool the foolproof AI", sort of like antivirus software vs computer viruses in our world.

The first AI targeting system will have a script telling it what to look for, probably heat signature to detect an engine and visual confirmation that it's an enemy vessel. Then someone will develop a decoy system that can avoid being targeted by the AI. Then the Ultratech Empire will integrate that decoy system into the AI's software so it's foolproof once again. Then the Space Pirates developing cloaking tech to make their vessels look like allied vessels (or an encrypted key, or whatever). And so on and so forth.

This would mean that, although AI targeting systems exist, only the richest and most powerful fleets will be able to develop / purchase an AI that is "smart" enough to take out the enemy. And only the finest space pirate crews will have the tools to outsmart them, even briefly, so that they can get in some hits or escape or whatever. But most crews will also have Human gunners, if not just Human gunners, because they're not AIs and will adapt to new decoys and maybe see through an enemy's tactics.

You can come up with any number of limitations for the AI to underperform, for example it needs a HUGE amount of processing power to react at the speeds necessary in deep space combat. Which means that "Gunner" becomes a real Human profession, with entire careers made from it.