Does the AI cheat or was this a bug?

Just for the record, I have noticed AI using PA from very odd angles, looks like they are equipped with gimbaled PA's.

I have also seen anacondas using iCutters as ammunition. Literally shooting cutters out and impaling this Fed 'vette with them.

Oh look, anecdotal daydreaming isn't that hard after all :D

I actually have more legit proof than you do as it happens...

3id5Wp.png
 

Well, I haven't had any yet and I spent my whole play session in a combat zone last night. As far as I know they don't have any, but if they do, then that's fine. They are not cheating. And that reply from support doesn't actually make it clear that they are back in.

Just for the record, I have noticed AI using PA from very odd angles, looks like they are equipped with gimbaled PA's.

This could be lag. Difficult to know.
 
Last edited:
NPCs carrying Low Temp Diamonds definitely do cheat - their FSD has no cooldown when interdicted - they start spooling up immediately.
 
I just got into a fight with a competent AI Viper. I'm in a Cobra MK III and I encountered the Viper at a wreck site. He scanned me looking to see if I had any cargo, then set about attacking me. So I deployed hardpoints and started shooting back.

I managed to strip his shields pretty quickly and got his hull down to 2%... This is when he suddenly hit me with a weapon that stripped my shields with one shot, knocked out my weapons with another and destroyed me with the third...

Is it a bug, or is it a normal tactic?

Also, if he had that kind of weaponry, why not use it from the start.

And... if he only had 2% hull, how come he had enough working systems to even use a weapon that powerful in the first place? Surely he would be suffering major system damage?

They do cheat, it is a bug, but it's "working as intended" ;)
 
I'm going to have to start asking for video proof of the AI cheating. It's been said time and time again that they do not. It's becoming a meme.
 
The only NPC that scares me is... NPC AspX with rail guns. They do wonders on me. Rapid fire with double shot?
Now, NPC Anacondas. Usually I dominate them till they are 30% hull and then they unleash hell and stuff on me...

Off to git gud...
 
I'm going to have to start asking for video proof of the AI cheating. It's been said time and time again that they do not. It's becoming a meme.

Well cheating AI is a good excuse if things go south in the RES. But I think the OP was the victim of a bug, or maybe he was sniped by a cheeky CMDR with long range hammers.
 
I'm going to have to start asking for video proof of the AI cheating. It's been said time and time again that they do not. It's becoming a meme.

They do cheat, but it's not in the way we usually see in posts here ie FAOFF, impossible to miss rails, impervious to heat, never ending SCBs/chaff etc.etc. No, it's the instantaneous reactions to all of a player's inputs; this means you cannot pull moves to fool them - like you can with players - which is highly unsatisfying in terms of gameplay. Ofc you can use this mechanic against them too, but that is pretty lame & just makes them even easier to kill!
 
The AI don't play by the same rules that govern us.

I've never understood why developers do that. Surely it's more difficult to write a whole new set of rules than just point their ability code to the player ones.

On the surface it may seem like it's easier to use the same player rules but in reality it can be a lot harder because you need to program in a lot of conditions that are common sense to the player. If you can skip over programming those common sense rules in favour of a "cheat" that in 90% of scenarios wouldn't even be noticed then you save yourself a lot of work and potentially considerable performance overhead. AI programming is probably one of the most difficult parts of game design if you want to offer any "believable" behaviours ... I'm yet to find any game that has offered a tangible improvement on AI when compared against the improvement of graphics on yesteryear. It feels like every year graphics comes leaps and bounds but AI remains relatively stagnant.

Elite Dangerous does really well though, the last big AI update was a huge improvement that I'm sad to say was short lived.
 
This sounds like you came close to a ground installation with some heavy skimmers. They carry some nasty missiles that will easily rip a cobra up in three shots.

People seem to be getting confused here. I was not near the surface, I was in a deep space USS with some debris.. That's why I said a wreck.

They do cheat, it is a bug, but it's "working as intended" ;)

It's not a bug, it's a feature... Lost track of how many times I've heard that excuse :)
 
Well cheating AI is a good excuse if things go south in the RES. But I think the OP was the victim of a bug, or maybe he was sniped by a cheeky CMDR with long range hammers.

100% sure there wasn't another CMDR in the area. Just me and the Viper and some bits of debris. But we wandered out of the debris field during the fight.
 
The Mistress has spoken!

AI does not cheat guys. I actually had a nice chat with Sarah at Lavecon where she's explained how they fly. they don't cheat :)

Alternately, the AI do things that they shouldn't and these things are unknown and unintentional to the programmer who wrote the code - i.e.... ... bugs! :)

Case in point? The period of uncertainty between the 2.1 AI going live, and people being unsure of whether the AI were just darn good at destroying ships quickly, or something else was going on. As we now know, the AI were messing up engineered weapon effects to give things like 'chain-gun' plasma accelerators. There was no intention for the AI to cheat, but they did so all the same. Ooops.

And the tactical ability for the AI to fly without cheating, is also separate from some of the AI RNG spawning effects - e.g. harassment of players by mission-spawned AI across systems that the AI couldn't have actually reached due to jump drive limitations.

*Edit* I'd have liked to have been a fly on the wall for your conversation about AI. For my sins I've been involved with a bit of agent-based modelling work over the past few years, and trying to model reactions to what their 'imperfect' (but actually perfect) senses are telling them to act 'more human', is a lot of fun. :)
 
Last edited:
Back
Top Bottom