#Maybe add the AI you were promising in the promotion material?

1 messages · Page 1 of 1 (latest)

plucky cedar
#

Listen, if someone is even gonna read this, its great and all that you expanded the game world and added more to explore and are constantly fixing bugs but what about the AI? What happened to the neural network you promised from the beginning and removed last second? Now I have some experience with machine learning and neural networks. For all we the community know there are a 100 different reasons why this self learning AI that was one of the MAIN selling points was scrapped but here's something you can try to implement a similar system. It's up to you if you want to listen ir not but here's my idea. Firstly, instead of the whole "neural networks that learns from the entire playerbase" thing you can do the processing locally on the player's computer for the neural network. to speed up the learning process, run a good few instances of the different neighbors at the same time that way when the player's in a chase it will reflect to all of the other instances like a mirror, making the learning and adapting process more efficient and faster. Secondly, if you do add the self learning system then you should add an option to turn it on or off for people with less powerful machines although most high quality computers that are made today should be able to handle the neural network. Thirdly, there are many many machine learning methods but the best for this case in my opinion is reinforcement learning, specifically deep reinforcement learning, maybe you know this but to be safe at its most brief overview its regular reinforcement learning but with a neural network that manages everything.

In conclusion please bring this feature to the game because it will add a huge amount of replay value and try to consider my ideas for just a bit. it will help to stich up this franchise's open wounds so people will actually say something good about it and will help your brand name in general.

lyric grove
#

Explain to me how the freaking Hello Neighbor Pre alpha had the best AI of any Hello Neighbor version ever

lunar heart
#

in short, the AIs and houses were made to work of each other

#

but when you mess one of them up or none of them reach a basic threshold, the AI won't work

#

Pre-Alpha (and to some extent Alpha 1) AI worked because it had everything it needed to beat you, and the house was made to work with the neighbor's AI

#

the following versions of the games changed something about the houses or the AI that it messed up the quality of the AIs (Like straight up not updating the AI to actually adapt to the new houses, Alpha 3 Style)

#

Ex: HN2 AI has the problem of it not getting enough information about your location, and the fact that some of the houses weren't really made with them in mind

plucky cedar
shy beacon
#

Yikes. Yeah the current HN2 AI is horrible.

plucky cedar
shy beacon
#

The AI in 2 is the main reason why I still prefer HN1

weary zinc
viscid fulcrum
#

he did confirm that the NN would be in HN2

#

but most likely he was doing that under the influence of tinybuild

plucky cedar
lunar heart
#

like FG HN2 Ted's house sucks

#

and an adaptive AI wouldn't really do much to fix the problem

viscid fulcrum
#

It was made pretty much perfect in patch 8

lunar heart
#

Honestly

#

I don't think so

#

for me, upper floor is still a safe heaven

#

Ted really can't do much if you just keep your distance

#

Bottom floor is fine tho

plucky cedar
lunar heart
plush marsh
# shy beacon The AI in 2 is the main reason why I still prefer HN1

Honestly, besides the glitches and camera spamming, I feel like Hn1s ai did a better job at keeping me on my toes than the hn2 ai. I feel like if they changed some things it would be alright. I think if there was some sort of combination of both games' ai it would be okay.

#

And with camera spamming, I think they could have it so that cameras could only be placed above or next to a door or on a table so that they can't just be placed every few inches on the wall

plucky cedar
#

they probably arent reading this post