They are beyond useless. Someone afked/dced out at start of my game and it was pointless to try and come back after it. We almost did it but its near impossible. Either make these bots better or don’t count those games to my overall score.
Lol, dc is always counted in every game.
The bots are 10 times better than what you get in LOL, where the dc is just in base and you are 4v5.
They promised it last Blizzcon I think. But you know, then they reduced the development team. So it might be out of scope now.
why would I care about LoL? I’m playing hots not lol for a reason
Because you must have a view of whats feasible and better bots is not feasible, its too complex to have AI be able to be good in Diamond games.
idk, I’ve seen AI beating top korean players in sc2, idk how hard would it be to make gold league level ones
try to define that term then 
If you let it run, chances are it would be placed in gold league I guess…It wouldn’t be game braking if you ask me to give us a teammate like that. anyways, some bots are much better than others, like kelthas or tyrande bot are awesome, but some are like headless chickens, randomly running around alone and hearthstoning, I don’t get it why. If levels are above 10 don’t let the bot soak, let it follow team.
Come to think of it, blizz is obsessed with anti cheating that it screws 90% of us non cheaters. I get it why they take my points, because otherwise one guy could “take one for the team”, but if you ask me, I’d say most people are honest but get shafted with random or rage dcs.
You cant be serious… Have you ever played sc2? The top difficulty AI get insane boosts to farm rates and discounts on cost. Basically they can produce 2-3 times bigger armies on equal bases hence they can beat even top players.
How exactly you want that in HOTS, you want the AI level 20 when everyone else is 15?
Oh and sc2 is a 1v1 game, in case you didn’t notice.
I hate when people have absolutely zero idea about how the real world works and suggest things that are not feasible as if that is easy. Go read a book about AI and design one yourself if its that easy.
You cant be serious that you never heard of machine learning and compare sc2 AI with what I want in the game? Besides, you’re focusing on the wrong part of my post, if you CANT DO IT DONT TAKE MY POINTS. I lost like 3 games in a row due to DCs and lags of teammates.
I hate it when people with low IQ like you argue with me on the internet, you piss me off with idiocy.
I work in Data Analytics within one of worlds largest tier 1 banks (£1.1+ trillion in assets). Some of my job (and wider industry required knowledge for a data person) is machine learning and AI, so yes I have a pretty good view of whats feasible given that I see attempts being made to use machine learning almost every day (in a company with literally trillions to back it up).
Prior to that I was advising a range of multibillion companies in the Data space so again, I do have the knowledge. I will not bore you with my degrees and other qualifications in the data space but I am sure even someone as thick as you gets the point; I get paid a lot because top tier businesses somehow believe that I know my stuff when it comes to data and analytics.
Cute how you try to insult my IQ. Pathetic attempt though as we both know that you are a little loser hidden behind the screen, probably never held a real job or delivered anything in his life. No vision of what cost to benefit may look like from blizzards side.
Here is one last lesson for you, if you want to achieve anything in life you need to know your place. Yours is low and therefore it is vital that you realise when you are inferior so that you can actually learn from people with more knowledge, experience and skill.
There is a team working on an AI driven by machine learning for DotA 2. They showcased their work on one of the championships in that game facing one of the teams. In order to make that match at least a bit competitive, they had to work with limited premade pool of heroes which both (AI and the real players) could pick from, otherwise there would be simply too much branches and possibilities. Even with limited hero pool of exactly 10 heroes (in 5v5 game) the AI was extremely predictable. It seemed like an equal match only in the first game, where the real players played really cautiously, but once they adapted to it, they steamrolled it in following matches. Also AI did very weird actions across the match like activating ulti to clear minion wave and so on (making commentators and people watching it laugh).
Surely, with larger data set, it could improve and be more reliable, but it wasn’t proved yet. Do you really think they are not trying to do something like that at Blizzard as well? They wanted to rework MM to analyse your gameplay and give/take you portion of rank points based on how you performed and also match you with similarly performing player. They were using machine learning to judge your gameplay. They put it on live server in xmas 2017 or something like that. It ended in a HUUUUGE fail. The system was completely unreliable and broke whole MM. Game was unplayable for a few weeks after which they just reverted whole system to previous state. They initially claimed that the problem was caused by unrelated issue and that they will put performance based MM (how they called it) back online soon, but after few months they confirmed that system is unreliable and is very slow to detect and adjust to meta shifts, but that they still want to use it at least to give players some feedback on what to improve. They still didn’t put it back even tho they claim the system already runs in the background internally.