[Goose]

Clearly they can be effective, as the video shows. The second shahed intercepted even looks to be a jet powered variant, and the drone still gets it. Note that article is from October- this is an area of active development, and we know Ukraine has been working on autonomous targeting as well.

Yes, improvements can be made in the process. However, as the article from October pointed out you need a radar, communications systems, etc. for each of the defended points. These things require money and organization which are not in abundant supply in Ukraine. These requirements do not change. The need for highly skilled pilots is at present a significant ongoing issue. AI may remove that constraint eventually, but it has not yet done so.
 
As for the interception of a jet powered drone, that also happens every day in Ukraine by machine gun fire too. If you have good geometry it is quite possible to collide a "slow" interceptor drone with a "faster" jet drone. You may need a head-on collision, but it can be done. That doesn't mean it is an "effective" weapon. It may be the best you have, but that may well not be enough.
 
Most importantly, no matter how much development goes on or how sophisticated your targeting software is, you cannot defeat geometry. If a Shahed 136 is flying directly towards the acquisition radar you have 8.3 minutes to launch your FPV drone and acquire the target visually (radar 10 feet in the air, drone at an altitude of 100 ft).  This is a "best case" scenario. You can defend yourself :-). However, if the drone's flight path is more than 16 miles either side of your radar, you don't detect the drone at all. You are safe, but the drone is on its way to its target. If the drone flight path point of closest approach is 15.9 miles from you and not closing with your position at that point, you have at most 1.9 minutes to launch and acquire. That isn't happening. The field of view of an FPV isn't all that wide. Acquiring the radar detected target at range (at night remember) isn't easy. If it were, a radar wouldn't be necessary. If you then have to "run down" the target, it can simply be impossible, depending on the geometry at that moment. A jet powered Geran 3 halves the time available to react.
 
All this aside, you would need radars and drone launchers/operators spaced every 30 miles (best case) or so along a "line of defense" to even detect all the drones, let alone shoot them down. Even then, two drones in the same "box" overloads the system unless you have multiple drone operators (and drones) at each site. Of course, on any given night most of these assets would be doing nothing. The resource consumption would be massive. There is simply no way such a system can "work". The geometry is just too difficult.
 
OTOH, an airborne radar at 30,000 ft will give you over 2 hours warning of the drone approach, assuming it is launched from more than 230 miles away. Not only that, the Shahed 136's ground track will be apparent after 10 or so minutes of flight. This allows interceptor drones to travel a significant distance before intercept. Such a system is much more difficult to overload. Such a  system can work. It is still expensive and complex, but it can work. A ground based one cannot.

[Goose]

Clearly they can be effective, as the video shows. The second shahed intercepted even looks to be a jet powered variant, and the drone still gets it. Note that article is from October- this is an area of active development, and we know Ukraine has been working on autonomous targeting as well.

Yes, improvements can be made in the process. However, as the article from October pointed out you need a radar, communications systems, etc. for each of the defended points. These things require money and organization which are not in abundant supply in Ukraine. These requirements do not change. The need for highly skilled pilots is at present a significant ongoing issue. AI may remove that constraint eventually, but it has not yet done so.
 
As for the interception of a jet powered drone, that also happens every day in Ukraine by machine gun fire too. If you have good geometry it is quite possible to collide a "slow" interceptor drone with a "faster" jet drone. You may need a head-on collision, but it can be done. That doesn't mean it is an "effective" weapon. It may be the best you have, but that may well not be enough.
 
Most importantly, no matter how much development goes on or how sophisticated your targeting software is, you cannot defeat geometry. If a Shahed 136 is flying directly towards the acquisition radar you have 8.3 minutes to launch your FPV drone and acquire the target visually (radar 10 feet in the air, drone at an altitude of 100 ft).  This is a "best case" scenario. You can defend yourself :-). However, if the drone's flight path is more than 16 miles either side of your radar, you don't detect the drone at all. You are safe, but the drone is on its way to its target. If the drone flight path point of closest approach is 15.9 miles from you and not closing with your position at that point, you have at most 1.9 minutes to launch and acquire. That isn't happening. The field of view of an FPV isn't all that wide. Acquiring the radar detected target at range (at night remember) isn't easy. If it were, a radar wouldn't be necessary. If you then have to "run down" the target, it can simply be impossible, depending on the geometry at that moment. A jet powered Geran 3 halves the time available to react.
 
All this aside, you would need radars and drone launchers/operators spaced every 30 miles (best case) or so along a "line of defense" to even detect all the drones, let alone shoot them down. Even then, two drones in the same "box" overloads the system unless you have multiple drone operators (and drones) at each site. Of course, on any given night most of these assets would be doing nothing. The resource consumption would be massive. There is simply no way such a system can "work". The geometry is just too difficult.
 
OTOH, an airborne radar at 30,000 ft will give you over 2 hours warning of the drone approach, assuming it is launched from more than 230 miles away. Not only that, the Shahed 136's ground track will be apparent after 10 or so minutes of flight. This allows interceptor drones to travel a significant distance before intercept. Such a system is much more difficult to overload. Such a  system can work. It is still expensive and complex, but it can work. A ground based one cannot.