Most drones should be part of the artillery, not the air force

Looking at the drone prototypes coming out of the USAF, VVS, and PLAAF, I see three main mission types for drones:

1) Observation/Recon

2) Air-to-Surface Precision Strike

3) Air-to-Air Combat

I order these three mission types in terms of cheapest to most expensive (on a per-unit basis). (Also ordered from easy to acquire, technologically, to requiring significant R&D).

If 1 and 2 are cheaper and require less technology than 3, then 1 and 2 systems will predominate in numbers and also arrive on the battlefield sooner.

The missions of 1 and 2 make much more sense for the artillery to handle than the air force. Yes, the air force should have drones--but they should be highly advanced air-to-air "swarmers" that can effectively contest air parity situations and clog up enemy air defense systems without risking valuable pilots.

Imagine a drone in the 3-5,000 lb payload class (analogous to the MQ-9 Reaper). Most of the time, it flies at 250-300 miles an hour, at an altitude of about 15-35,000 feet. It does not have a long-range air-to-air sensor suite or any EW capacity, or any ability, really, to engage in high-g air-combat maneuvering.

It makes zero sense to have this style of weapon in the command of a pilot. It's not even a CAS-style aircraft, designed for sortieing in and out to ground-designated targets--it's a loiterer, a patient "stalker" that sits there until ground forces or its own sensors have ID'd a target of opportunity.

On the modern battlefield, then, this type of drone operates in a fire support role, because of its ability to loiter. If you want to talk about using these drones to replace something, use them to replace mortars and field guns, because with enough cheap precision strike drones you can entirely dispense with costly, large, and easily targettable artillery batteries.

Imagine an army, where every company commander has 10+ Hellfire- or Atava-style missiles at his disposal. Or, if he needs to perform anti-personnel suppression, 10+ Hellfires can become 50 or so RPO-Shmel thermobaric rockets. What's more, these drones can be kept fueled and ready to go on tracked or wheeled carriers, ready to be launched to provide instant, lingering, and sustaining fire support with a CEP of about 5 meters. Whenever the drone runs out of ammo, it just lands by the control truck, gets outfitted with a new batch of ordnance, and up it goes, ready to rain death. Why would he ever need mortars or 152mm guns?

He can operate completely detached from brigade- or even battalion-level fire support. Why? Because if all his fire support now has a CEP of 5m, this can let him have a near 80 or 90% hit rate, which means that you no longer need to dump tons and tons of shells on an area in order to hit it. This lightens the logistics load.

Alternatively, using even lighter drones with just observation/targetting abilities lets infantry or armor act as its own forward observer. Simply put an infrared targeting laser atop the drone and you can guide pinpoint laser-guided artillery, rocket, or missile strikes from higher level fire support assets without needing special training; plus you can do this pinpointing from an overhead angle, which allows for a far greater level of accuracy.

However, putting these drones in the hands of air force pilots takes away this sort of flexibility, and the most important advantage of such an operating model: localizing the C4I load. With drone-based fire support under the command of company or even platoon-level team members, small units can call down a whole lot more precision strike ordnance without needing to contact battalion or brigade HQ. This drastically shortens the OODA loop and dramatically improves the efficiency of small unit leaders out in the field.

The Pentagon is already too far gone down the road of an Air Force-centric drone warfare model, but I think that the PLA still has room to properly develop its drone doctrine into something that fits with the requirements of modern war.

Thoughts?

I would, quiet frankly, disagree with that.
Reaper class drones aren't just larger RC toy planes. They're pretty bigflying machines. So for one thing, you won't be able to carry them around on vehicles and just launch them. Even MQ-1s need a fixed base, semi prepared, (field) base to be operated properly. And you can't just have random guys in arty batteries have these things flying around. To use those drones effectively in a possibly congested airspace requires some good airspace management and operators who understand that and how to move airplanes around in a coordinated fashion. That is some proper air operations. Plus, Predetor sized UAVs aren't that cheap either, so I guess even the PLA won't be in a position to issue them to regular arty units throught.
Now Mini UAVs are a different matter, those have been issued to arty units before, including in the US. They can act effectively as a means of supplementing a forward observer. Have these assets then control Mortar / Arty fire, maybe even precision guided projectiles with an onboard laser designator, is a much more viable path, IMO.

The unarmed reconnaisance will likely disappear with reconnaisance having at least some rocket powered grenades for short OODA-loops on soft targets. Heavy armament on drones is only cost efficient for low force density environments and very fast strikes. The closer a drone is to a significant number of own troops, the cheaper is reliance on ground vehicle based mortars and rockets and combinations of these.
Lightly armed reconnaisance drones can circle over the heads of ground forces with light and heavy artillery.

Heavier drones with heavy armament cruise over longer distances searching for targets, but their use has overlap with manned ground attack crafts. They have the longer endurance and unendangered pilot as a bonus over manned systems. Their main advantages are the legal loophole and roughing up armed low-tech militias.
Much of their mission on land can be carried out for a fraction of the costs and requirements of trained manpower in aircraft maintenance by arming squadrons with a light artillery piece (combing mortar and howitzer) and a cheap reconnaisance craft on their patrols. The killing without boots on the ground is open to all kinds of manipulation from current propaganda to constantly hitting inflated rubber tanks.
At sea, these heavy armed drones are of much more use. They will have a major impact on sea control and strike range with targets being much harder to fake. The problem is radio-silence or a non-detectable way of communication. Some of these drones at sea can be used for niche applications on land. That's why I would favour to make them a domain of the navy/marines, who can support their brothers in arms for the niche applications far from the sea. Small to medium sized carriers will have a major benefit in sea control capability from these aircrafts.

Air-to-air combat drones are more about signal intelligence and electronic warfare. Electronic warfare can at the same time serve as a radar source (see the latest developments of "passive" radar). The aircrafts are used to maintain an information flow, they do support units on ground (that can fly jets or missiles) by providing information for strikes. As armament, long distance missiles would be suitable, while short distance armament makes little sense, because you can't have high endurance with great agility. The long distance missiles from these drones can attack flying enemy information infrastructure and provide more information on possible contacts (with the ability to land intact and be retrieved) without giving away much own information for counterstrikes.

Another possible application is in mixed manned and unmanned systems with unmanned aircrafts of the same specifications serving as wingmen to manned follow-me aircrafts.This can double numbers without doubling training costs. In air combat, such a wingman can operate as a programmed support for the pilot (and each pilot should program his personal wingman during training sessions) or be replaced by an ace pilot in remote control. The flying pilot and the remote ace pilot must have trained together and the flying pilot must be able to shut off all remote access allowance in order to defend against hacking. The net result would be better human resource management and an aircraft closely (short distance, hard to jam and detect, no time lag) connected to a human decision maker that can commit to continuous human-disorienting maneuvers at high g.

Retrieveable cruise missiles can blur the distinction from drones, but they do have a major size increase that limits the amount of munitions carried. While feasible, they might not be the direction followed, because there are more cheaper options like forwarding observers for strike missiles with rocket engines. These observers can be armed in order to supress countermeasures against the main attack and would be the aforementioned heavy ground attack drones. These unmanned aircrafts are built for endurance with much slower speeds than the strike missiles that fly at several times their speed. A drone flying at mach 0.5 would direct missiles with mach 3.0 with a strike range of 2,400 km. In the time such a drone moves 400km from 1,900km(seeing a target 500km away) to 2,300km range (100km away from the target), the missiles cover all the distance to the target.

That can be the crucial point about drones, make them capable not only to observe, but to direct fire and supress efficiency of countermeasures against incoming surface-to-surface strikes. Supressing countermeasures will include hard- and softkill armament. Drones do have the capability to fly longer range and endurance missions than manned aircrafts. BVut drones need to get rid of a chatting communication that limits them to "counterinsurgency" for this application. Such drones need a counterpart that can cost effective deliver a strike of suitable size over that distance. Rocket engines will have an efficiency and range increase by giving them scramjet features. This idea would not replace manned systems, but it does provide an alternative for the use of massed complex attacks. The scouting drone has a major cost benefit by providing an observer system to enhance less developed and expensive targeting sytems on the missiles.

The old parasite carrier idea might be worth another try with unmanned systems that fly at slow speeds. A blimp could serve as a docking station for unmanned cruising drones with heavy armament that can be rearmed and refueled with part automatic and remote controlled maintenance.
It's not about having lots of strike crafts, but a good area coverage at low cost. A slow moving blimp by far outruns all vessels at sea and land and is even capable of carrying a missile strike package that has a range bonus from being air launched, while at the same time no aircraft beats it in endurance. A blimp also has a very large structure for use as antennas. Such an unmanned system is ideally suited to run operations over a large area of low conflict intensity and does increase the range of centralized facilities(=fewer and thus cheaper total infrastructure costs) on ground that do the major overhauls of the unmanned systems.
Manned bombers (from large to small as COIN aircrafts) do have an application in this environment for a strike complex against stationary targets that are either too far away for a direct strike and must be reached by aerial refueling or not worth the expenditure of sending missiles.

Summary:
I still see drones as complementing other weapon platforms and enhancing their cost effiency in an organic development, not a revolutionary replacement. They will alter the way traditional platforms are employed.

The unmanned systems increase observation capability in an environment that is heavily reliant on avoiding observation in order to survive the existing strike capabilities.

Drones will have a massive impact on observation and strike range from other platforms, but themselves only carry very limited ordnance, akin to snipers.

Fighter jets can spread thinner with half of them being unmanned wing"men".

All ground forces to the lowest level will have unmanned aerial support constantly with them that does not only observe, but strike and help to direct strikes of ground based artillery that will be as common as machine guns today. These ground forces do include air defence forces with missiles and fighters.

Massed airstrikes against high value targets will increasingly be conducted by a combination of observing, directing and supressing drones with surface-to-surface missiles and not bombers or fighters. That makes drones akin to stormtroppers, who advance prior to the main assault and hold pockets that disable much enemy resistance.

Low force density environments will be dominated by unmanned systems and, as outlined above, concentration of manned systems can decrease due to the availability of unmanned systems for different strike models. Thining out does help against the devastation due to observation.

This leads to exactly the kind of environment where "sniping" with drones can be more important, as long as none invents a mobile strike complex that is a concentration capable of defending against all kinds of threats within its battlespace around the vulnerable core.

First The USAF is not the only service in the US militaty too use UAV's. They all do The Army, Navy, Marines and Air force. Second limiting UAV's too Artillery would be very short sighted.
There are and Should remain and be pointed to for this case Many different classes of UAV's, Each has a nitch too fill and a Unique set of Pro's and Con's. I am going too try and Clarify this.

Small Man portable systems. AKA Micro Air Vehicles
designs like the Honeywell RQ-16A T-Hawk, Dragon Eye UAV, BirdEye 100 Backpackable UAV and even smaller units Some designs even fit in the hand. these are for use by infantry they enable a birds eye view of possible trouble. there is development on going but Ideas such as arming them with small arms are still troubled because of there size and payload. very very short range and have short loiter times Easily . These Drone types should be issued force wide too infantry, SF and even Law enforcement.

Short Range long endurance.
Like the RQ-7 Shadow. These system best Suited too the Artillery base and mounted vehicles they are larger have more sophisticated sensors and payloads There is now the definite capability of even arming them with small weapons.
These Craft are best suited too Artillery units because they have the range too sea over the Breath of cannon Range as well as the loiter Time too see the Before and After Effects. In the First Gulf War in, the battleship USS Missouri Used A UAV of this Class too target Iraqi Positions and then assess the Destruction post firing. Iraqi Forces in the Range of Mighty Mo's Guns Quickly Learned the Sound of the Rq 2 Pioneer and Began surrendering too it knowing that if they did not vacate or give up the Battle ship was going too soon be sending really really big Shells.

Medium Range Tactical UAV
Predator, Reaper and others. These are getting all the Glory these days they deploy a limited number of weapons like Hellfire missiles. Roughly the size of a WW2 era Fighter they are often Turboprop although some are Helicopters and sometimes Jets. they offer tactical overview and allow smaller scale options for known trouble spots.

Extended Range
Global Hawk, Soar Dragon and others. Strategic options akin too the U2 they are large Jets used too loiter for long times and when needed penetrate too see what is happening. Often unarmed they are oft the size of fighters. There are also some that are very very low speed ultra long endurance Basically Stealthy spy Satellites with wings. these are Strategic types meant too give the birds eye of the Birds Eyes

Unmanned Combat Air Vehicle
Strike bombers who left there pilots behind.

Spy satellites
Gods Eye view

other
Use your Imagination.