TL9 Heavy Tank

[Nightrider_88]

IIRC, one crewman was able to operate Strv.103.

[Curmudgeon]

Originally Posted by Rupert

That's in large part because of the force imbalance there, though. I mean, about the only thing that could've gone worse would've been if the landings had been on the wrong island.

Not entirely. If the U.S. hadn't been as swift in exiting as it was, or if things had gone a bit swifter in other quarters, it could have gone very wrong. Both the UK parliament and the Canadian parliament had motions on the order paper to consider a military response to the invasion of Grenada. That's a military response on Grenada's side. It wouldn't surprise me if there weren't similar motions scheduled in the other Commonwealth Realm parliaments. And if that happened, a Commonwealth-wide response might be forthcoming later. As it happened, the U.S. left before the motions came before parliament, but things could have gone much worse than just landing on the wrong island.

[ehrbar]

Originally Posted by Curmudgeon

And if that happened, a Commonwealth-wide response might be forthcoming later.

Even in some weird scenario where Margaret Thatcher decided to pursue such a response, or was removed by a just-elected huge Conservative majority and the new PM decided to do so, (or perhaps we even step back a few months and magically have Michael Foot win the UK election,) a "Commonwealth-wide response" still isn't going to happen.

Why? For the simple reason that a number of Grenada's Commonwealth neighbors (Antigua and Barbuda, Barbados, Dominica, Jamaica, St. Kitts and Nevis, St. Lucia, and St. Vincent and the Grenadines) both requested and participated in the US intervention (with the 350 troops of the "Caribbean Peace Force" landing five hours after the first US forces).

Even with a rather more botched intervention, the participation of those seven in the intervention throws a stick in the spokes of any attempt to organize a "Commonwealth-wide response".

[Sam Baughn]

Originally Posted by Rupert

Good controls for operating a gun are not the same as good controls for driving a tank, especially once you've made them simple and (theoretically) soldier-proof.

At TL9, I imagine most 'controls' (especially weapon controls) are likely to be 'tell the AI what to do'. The best interface for 'designate route to drive along' is unlikely to be the same as the best interface for 'designate which targets to kill' but it might be closer than manual steering and gunnery controls.

[Rupert]

AS I think I've said many times, here and elsewhere, I'm fairly sceptical about TL9 'AI' driving combat vehicles. Shooting? Sure, once the commander has designated something as a target or the AI's identified it as a target, but driving? Not so much.

[johndallman]

Originally Posted by Rupert

As I think I've said many times, here and elsewhere, I'm fairly sceptical about TL9 'AI' driving combat vehicles.

Yup. Driving is currently unreliable on roads. Off-road, in a vehicle capable of busting through many things, but vulnerable to stuff getting tangled in treads, is a whole different set of challenges. And a system that can do that has to be constrainable to road behaviour for peacetime and friendly territory.

[Polydamas]

Originally Posted by johndallman

Yup. Driving is currently unreliable on roads. Off-road, in a vehicle capable of busting through many things, but vulnerable to stuff getting tangled in treads, is a whole different set of challenges. And a system that can do that has to be constrainable to road behaviour for peacetime and friendly territory.

And friendly infantry will be camouflaged and concealed because, you know, if a friendly computer can easily spot them and know not to drive over them or get between them and a target, an unfriendly computer can easily spot them and send something nasty their way. In many wars both sides use the same kit with just small markings to tell them apart! Expecting a mid-21st-century AFV to drive itself in complex combat environments sounds very ambitious to me.

Human beings are terrible at switching from passively observing to taking charge in an emergency. AFVs in combat encounter more emergencies than a delivery truck carrying rebar to a potash mine. And because an AI can not be held responsible, an AI can never make a management decision. That goes double for decisions like "drive over this wrecked car which might have civilians in it, or go around the block losing time and possibly exposing our side armour."

Edit: AVFs have a proven technology for self-driving long routine journeys. Its called "loading them on trains."

[Witchking]

Originally Posted by Polydamas

Edit: AVFs have a proven technology for self-driving long routine journeys. Its called "loading them on trains."

Of course as the Russians are currently finding out; that too has unresolved issues.

[Polydamas]

Originally Posted by Witchking

Of course as the Russians are currently finding out; that too has unresolved issues.

I don't think there has ever been a state which could move thousands of AFVs to Ukraine without extensive use of trains. Driving tracked AFVs for thousands of kilometers causes breakdowns and destroys the roads, and flying them all in is too expensive. Landing them in ports is vulnerable to attack too (and then they still have to move to the front or the training areas).

[Ulzgoroth]

Originally Posted by Rupert

AS I think I've said many times, here and elsewhere, I'm fairly sceptical about TL9 'AI' driving combat vehicles. Shooting? Sure, once the commander has designated something as a target or the AI's identified it as a target, but driving? Not so much.

TL9 AI isn't bad at perception and world-modeling tasks the way real AI has been.