devtrackers.gg
Just in case any Rioter reads the comments: why did the bots in intermediate coop suddenly stop having skins?
Up until a few months ago, bots use to have skins, and it was often a cool way to see new skins in-game.
Just in case any Rioter reads the comments: why did the bots in intermediate coop suddenly stop having skins?
Up until a few months ago, bots use to have skins, and it was often a cool way to see new skins in-game.
Unsure, but will pass the query on.
Honestly, long needed. I've tried to introduce friends to the game with bots since they were nervous about starting against actual humans when they didn't know what they were doing, but the current bots are such a far cry from competitive that they couldn't beat Iron IV players.
Hopefully this doesn't level up the "other" bots, the leveling scourge that sits in aram/low level rift games wasting people's time.
Yeah, we very much need a better, more tailored to player needs range of bots to create a better onboarding experience. It's a tough game to get into, and right now we see a lot of new players drop off when trying to make the jump from the tutorial to Co-Op versus AI and then again when trying to go from Co-Op versus AI to PVP. Being able to give players more gradual ramp ups in difficulty early in their experience would do a like to help with new player retention, which is really important for games like LoL to continue to thrive long term.
In terms of impact on levelling bots - no, wouldn't do anything to enable those. We did a ban wave recently that hit a lot of those accounts, we've also got some further work underway that should make leveling bots, scripting etc easier to detect/quicker to ban too. We generally avoid talking about the details, exact timings etc on that stuff though to avoid giving bot/script creators any information unnecessarily.
Who cares?
Bots are training wheels and you shouldn't really be needing them long term
New players are our initial focus with this bot upgrade work. While most players likely won't be interested in playing against bots long term a good onboarding experience is really important to help long lived games like LoL maintain a thriving community long term.
Honestly, I played over 1500 games with and aginst bots so I can't wait for this!!! I hope we will also be able to get more points for passes playing vs AI
Hey! Thanks for being super excited for our bots! We are super excited to make them feel great from newbie to co-op versus AI mains. :)
As someone who played an entire year of bots before normals and spent hundred of custom only against bots this could be interesting
I hope they make it challenging enough that casuals are able to play it often and long enough as a side game
100%! I played bots until level 17 and only through sheer willpower (and the power of friendship) did I stick around long enough to learn to love League, and now it's the major reason I joined Riot.
We have a long term goal to provide challenging bots for all skill levels, since even now when I want to play, I usually want to try something fun and see if it would work out at my skill level. I think bots is an excellent opportunity to meet these types of needs. :D
Does this mean Renekton bot going afk will be fixed?
Yes! We are building a much more scalable and maintainable system to ensure these kind of bugs are easy to triage and fix.
Same, I like playing this way and can;t wait for the new bots!
Please try our bots out when they come out later this year for the PBE test! We are currently aiming for helping newbies, but we do have much longer term goals that your opinion would help us influence!
Just in case any Rioter reads the comments: why did the bots in intermediate coop suddenly stop having skins?
Up until a few months ago, bots use to have skins, and it was often a cool way to see new skins in-game.
Unsure but thank you for flagging! We do have this as a Must requirement though for our new bots, so thank you for saying you appreciate it as it makes it an even stronger Must when we launch. :)
It won't be a good training system though, the only part missing is jungling and wasting resources on the bots isn't going to improve that, for players to learn they need to play against players, pure and simple
We are adding in jungle bots :)
As an experienced player, would it be possible to implement a more advanced training stage, like a jungle gym in prac tool? Moving targets of different speeds, or attaching items to a target to test functionality as a solo player would be nice
We are building this system to be extendable to these goals and use cases!
Since this is new infrastructure, we are starting easy and scaling to other use cases once we are confident in the infrastructure.
With the addition of jungle bots, and I assume wanting new players to jungle, will co-op vs AI be moving to Quick Play from Blind Pick, whenever that drops
We are currently working closely with the quick play team to see what the new player experience should be, whether blind pick or quick play, but our bots are being built with both routes in mind.
For the 2 week on test though, we are aiming for blind pick and when the bots are ready to roll out on production, we will adapt them to the strategy of champ select at that time.
Is there any chance that Doom Bots will come back in the future following this change? Maybe as a higher 'challenge' difficulty for the new bots?
Let's just say I'm a very big fan of Doombots of Doom and want to bring back the memes of surviving Annie bot 👀
I remember my first normals game after I gathered up enough courage after crushing my bot games. I got completely destroyed. Glad they are expanding on the bots.
This happened to me too! My friends just told me I have to lose a bunch until I "got it" and it was super painful. We are specifically trying to address this as our first major usecase so that players have a much easier ramp into PvP
Sometimes I want to play league but don't have the mental energy to play vs real people. Bots are very useful for that.
Totally agree! I'd be much more willing to jump on (and convince my friends to jump on) if we had a way to avoid incurring the mental energy of PvP. (Or if we could just get 5 on to try something really silly, just to have fun).
Right but a tutorial is going to be vastly better than your bots, you know it, i know it, everyone else knows it
Agree to disagree on that! But regardless, a good tutorial also needs good bot tech to function.
Long ago, I've groups of friends that'd exclusively play AI intermediate and avoid norms/pvp
Is this still relevant? Is there any statistics of how many legitimate AI mode players are out there, just curious since this topic was never really brought up since forever ago
There is some data, but we are working to gather a deeper analysis overall around these types of players, especially once we begin to test them out on live! I know I personally would play a lot more if I had an equitable challenge in co-op versus AI that gives me the same skill expression without the worry of a PvP experience.
Overall though, my gut feeling says if we do make overall better bots, we can also draw a lot more players to play in our co-op versus AI modes (more folks like your friends!) As the current bots get really repetitive and don't provide the variety that PvP experiences offer.
Have you guys looked into deep learning applications for Bots? I know OpenAI was using an unstructured learning approach to train a team of 5 AI how to play DOTA and apparently they found relatively good success. I wonder if, with a well structured approach, and access to the LoL API, you could teach/develop macro sense in bots to a similar extent and with relatively low computational “intensity”. It would be really cool to see.
We are highly aware of this work in particular and are connected with a few folks over in OpenAI. We are also keeping a strong eye on the work around Juewu has been doing in the space!
A reason we decided to pivot our technical infrastructure is to hopefully allow a future to be inclusive of a variety of bots that provide challenging and fun experiences to a wide variety of player motivations and needs. It is one of the reasons we want to take our time to ensure our structure is sound enough for these opportunities
And while you level up your bots, could you please permaban all level 30 bought accounts raised by bot networks?
Buying accounts is by your own ToS' ruling illegal, yet every game in Plat upwards features some 'mysterious' prodigy with a level 30 account who gets 40 LP each win, is extremely toxic and plays way out of the league they're in.
And no, don't bullshit me with this "level 30s aren't all bad" nonsense. The level that these alleged 'new' players play on is absolutely ridiculous and you can tell from a single game they play whether or not they're new or an illegally bought account.
Also, their name history which I'm sure Riot can check would further drive the nail into the coffin.
Hello! Although I am only the PO for League Bots, if you check out our dev blog they talk about starting to do some ban waves around the bad kind of bots so hopefully you will see less of these kind of bad bots in the future!
would it be possible in a future to have a bot replace an afk player in normal games?
We are building toward a future where we have the capabilities to do this! Currently, we are working on the core infrastructure, and certain champs could be fairly difficult, but this is something we are jazzed to investigate once we have a stable and working system in place. :)
I am SO excited about this! Kinda sad I went for bot blaster at the end of last year instead of waiting 😅. This is so exciting, especially since PvE was mentioned a few times, and I never played back when PvE modes existed. My one wish is to experience updated doombots since videos made it look chaotic and fun.
I am also a big fan of the idea of DoomBots making a come back :)
Based Riot Teemo flair
I chose it for that extra chaos energy B)
Could you make it so bots obtain kill gold? I was sad to learn a “fed” bot is still on a consistent timeline for item purchases.
We are building our bots to play as players, so they will get the money and gold that players get (including kill gold!)
DOOM BOTS DOOM BOTS DOOM BOTS
DOOM BOTS DOOM BOTS DOOM BOTS
so maybe by next year you’ll have the beta released on live servers?
that’s the takeaway from “later this year we will release our simple beta for 2 weeks on PBE to gather feedback”
Thank you for commenting! I get the frustration as waiting is a pain, but we are also building out the core infrastructure and tooling around bots as well.
We want to launch as soon as we are able to, but we want to make sure that we can long-term sustainably improve the bots, maintain them, and expand them to other usecases. (For example, future game modes, replacing AFK players, making bots come out alongside new champs, new AI for mobs and jungle monsters, and fixing bugs quickly and easily).
All this infrastructure takes time and deep consideration for the impact on a bunch of teams, but I believe it will be worth the wait for the long-term happiness of our players.
hello I might be too late to ask this but will bots still be available on custom modes? Sometimes when I'm learning or relearning a champ I go to play custom 1v5 intermediate bots to get the vibe going. Hopefully this feature stays.
Our goal is to continue to support all current usecases of bots with our new bots!
It might just be a bit before we get them into a stable place where we can focus on other modes, but customs and practice tool are areas we plan to support! :)
Programming scalable bots for such a complicated game sounds like an insane undertaking, so much respect for all the hard work that is put into this. The leap between bots to players has always been a pain for anyone I know that is interested in League and imo it's the steepest skill check in the game so I'm glad it's being worked on!
Thank you! I shared this with the team and they are really appreciative. It is a big undertaking, but we are all very passionate about the work and are excited about the cool things we can do with it long term!
Hey Dashi, I was wondering, will some of this bot tech also be used to update AI for pets/camps/minions ?
Yes! We are building our tools to be extendable to other teams so that they can in the future. Our teams have chatted a few times about the possibilities!
This comment will probably get buried, but to any Rioter that sees this, have you thought about using machine learning to train bots? Or are you already using machine learning to level up these bots? I imagine that if you want to produce truly dynamic, competitive bots, a neural network trained on real games is the only way to do it.
You definitely have the data sets to train on, and you can even train on different data sets for different bot levels, i.e., "bronze" bots would be trained only on games with bronze players, "silver" bots would be trained only on games with silver players, theoretically all the way up to challenger. Honestly that sounds like an incredibly fun project and it would give players a wide range of bot levels to choose from. These bots would also automatically adjust to new metas if you simply retrained the bots every six months or so on new games.
We have chatted about it! A lot of the work we are doing will enable some potential in the future. We see these bots as a foundational step towards the possibility, as if we have good bot tech now, there is always a strong fall back system that still feels good to play against. :)
I don’t think anyone will see this but I HAVE to share
My dad is very caring and really wants to have interest in all of his kids interests. So when I became obsessed with league he immediately joined (especially since he never enjoyed FPS games and I came from Call of Duty to Overwatch to League. So when he saw me playing league he was like I gotta try)
After messing around with what he might like to play and stuff, he finally settled on Brand jungle. It’s ALL he wants to play, he loves him. Except when goes to play with humans he constantly gets flamed and is quite a fragile man lol so he doesn’t like playing with humans anymore. For over a year now all he does is play like two games of Brand Jungle a week against bots while I watch him and teach him (I’m a jungler too so that helps)
I can not wait to get home from work and tell him about this. He’s going to be so excited that he can be more competitive with bots
I really appreciate this, I'm going to share this with my team because it's so sweet!
We are really excited to have y'all try them out. :)
Usually those kinds of ai I've seen tend to win by having flawless micro(they "know" their range with pixel precision, their exact damage output, etc).
But when we look at the macro, and things we associate with thinking and smartness, we see a bleaker picture : they make a lot of mistakes and struggle to adapt their strategy.
How is the research progress regarding AI long term planning and decision making?
We are actually building in a team layer of AI that is essentially the macro, and it can influence the micro of the champion AI. This layer will not have cheats but access to utility functions ("How much approximate damage can be done in the next 2 seconds?") and a memory bank ("their last known position was X"), somewhat similar to humans where you go "he likely backed to base since I haven't seen him in X time and mid isn't pushed up, so probably not ganking mid, and we have a ward on Shelly so they aren't taking that" which we can fudge read accuracy at both the team AI layer and champ layer (like we make an error role that can be configured for different skill levels /types of play).
Right now we are just starting on the foundational bits of this layer, but we want to test this one out during our 2 weeks on PBE test!
We feel famous that a riot dev commented on this lol. He wants me to ask if you guys can buff brand jungle please, I told him I’d ask😂
Unfortunately I'm not the balance team but I'll make it more viable by giving him a more League-like bot game experience 🙈
Dashi if you guys do this right (the upper tier of AI at the end of your plan)….the coop vs AI que could end up being better bigger and harder than the pvp que 😂.
Few of us only hope that the ML AI are available in custom matches. A lot of us roam there instead of with randoms, even if just duo!
Late to the party I know, but this is insanely interesting, cannot wait for your pbe test!
That would be amazing to be honest! I would love to play against difficult bots.
I will say, one of our goals is to deprecate and replace all current bot features with our bots (so at some point we want to enable our new bots on customs!) but we need to get the bots to feel great first and then those areas are easier to extend to.
Please do try them out when we hit PBE and let us know how we can further improve them! Also, let us know of any bugs or weird behaviors!
I wanted to know: Though intro bots are coming first on PBE, is the plan for all difficulties to ship together on live?
Cause I feel like easy bots are probably the most important, you only play a few games on intro but then you need to practice a lot against easy bots before moving to medium and/or normals. It's also the best place to test champions.
Yea, the goal right now is to ensure intro has no major bugs/systematic pain points (especially at scale) given we are using fairly new tech.
Once that is confirmed, the intro bots feel good, we will go heads down for a bit to build out beginner/advanced queues.
Then when all 3 are in a polished-ish state, we would want to roll all 3 queues out to live environments and iterate afterwards to make them more dynamic/improved to higher skill levels/expanded feature sets.
Not criticising the approach (I'm sure there are lots of positives), but does that mean that you'll be in a spot where intro bots are ready to hit live but they might need to wait a good few months or even more to go live as they need to wait for beginner and advanced to be done? How does your team feel about potentially having to sit on great new intro bots while they work on the other difficulties?
Yea, no worries!
So for context, we as a team are focused on making it easier for players to enter (or get back into) League after a break by providing them a lower-stress queue/learner's queue with Bot experience similar to normal PVP. (Example: Arena is a cool new game mode, lots of players came back to try it out, and I bet more players would have played longer if they could practice and learn the new mode/ try different champ comps/augments/strategies against bots first prior to going directly into PVP, where they might lose the first 10 games before 'finding the fun'.)
With this goal in mind to help ramp players into PVP, if we were to launch our Intro bots now, the New Player experience would be "enjoy these new good Intro bots" to "Pretty bad Beginner and Advanced Bots" where you once again lose ganking, jungling, and objective training. -- Thus not hitting our true goal/intent as a team of making players have a way to feel confident to go into PVP.
Additionally, having a live product would mean our ability to make new features goes down pretty heavily as now we would be maintaining the live experience at a global scale, alongside building out those new queues -- thereby delaying those two queues much longer than if we were to launch all 3 at the same time.
That said, please do try to play on PBE when they are out as the more we can test our bots at scale and ensure a good player experience!