Human Behavior, as Guided by Artificial Intelligence

Human Behavior, as Guided by Artificial Intelligence

Reading Time: 2 minutes

Human Behavior, as Guided by Artificial Intelligence

This was a fascinating piece to write. It’s about neuroscientist Jeffrey Lin, who is dramatically reducing people’s toxic behavior in online gaming at Riot Games – by using artificial intelligence. This isn’t some future vision. They flipped the switch a few months ago and it’s working. 

What Riot has built is an artificial intelligence system that automates responses to toxic behavior in its flagship game, League of Legends. Think of it as an artificial immune response system.


Though I’m not personally a gamer, I felt compelled to understand how this system came to be. I wanted to understand how it worked and what its implications might be for influencing human behavior in other contexts. 

There are so many interesting facets to this story. I find it frightening in some ways, and very inspiring in others. 

Riot is clearly out front in applying artificial intelligence in gaming. I predict the next large-scale applications of crowdsourced artificial intelligence will be in social networks. Facebook and Google have snapped up the leaders of a particularly promising type of machine learning called Deep Learning, and Twitter is ramping up its artificial intelligence investments too. Demis Hassabis, CEO of Google’s DeepMind acquisition, recently noted that “In six months to a year’s time we’ll start seeing some aspects of what we’re doing embedded in Google Plus, natural language and maybe some recommendation systems.”

Hassabis made that comment about artificial intelligence here in Google+ just about six months ago. What do you think he was hinting at? Will it be used to better understand and match people’s interests here? Also, Google+ is pretty tame on the toxic behavior bar when compared to sites like YouTube, reddit, 4chan, and online gaming, but could something like this be applied here too? Do we need it? Would we want it?

I will probably end up diving back into this story again soon. In the meantime, I hope you enjoy reading it as much as I did writing it.

#humanbehavior   #ai   #artificialintelligence   #leagueoflegends

No comments

  1. I’ve notified just about a dozen people who I thought might find this topic interesting. If you are one of those people and you’d like me to not do that in future, please just plus this comment. 

    If you didn’t receive a notice, and would like to for similar posts in the future, then you can also plus this comment and I’ll do my best to do so.

    (HT to Denis Wallez for this technique)

  2. This looks really good. Thanks for the tag, Gideon Rosenblatt .

    I’m on a working trip for the next 6-7 days, hopefully I’ll find time to delve into this more deeply. 

  3. I liked LoL – though I haven’t touched it since my graphics card burnt out. I really like they way they handled F2P.

  4. Gideon Rosenblatt​ This looks fascinating, going to archive it to read tomorrow.

  5. Because I’m not much of a gamer myself, Grizwald Grim, and really wasn’t familiar with LoL, it was harder for me to really get the feel for what was going on. Lots of cultural references here that I just didn’t get. The “gg – easy” thing, for example. 

    Riot sounds like a good company though. It’s interesting to see the organizational innovation going on in gaming – Valve too. 

  6. GG is ‘good game’ – if used prematurely, before the game ends – it’s a little insulting. Saying that it was also ‘easy’ essentially implies your opponent is horrible at the game 🙂

  7. Gaming is a really strong culture, Grizwald Grim​. I’m amazed at the nuances, and the gg thing is a great, subtle example. Same with this place. G+ definitely has its own culture, distinct from Twitter and FB. To think that an AI might pick up on that is fascinating.

  8. How did the AI persuade Riot to allow it control the people?

  9. I think it asked nicely, Sakari Maaranen. 😉

  10. good morning friend 

    how are you 


  11. Here’s some more thoughts:

    When are we going to figure out using the term ‘artificial’ is a species-level manifestation of what we most commonly recognize as racism?

    I finally caught Ex Machina this week. When does AI start qualifying as ‘people’?

    If corporations are people, shouldn’t we take a closer look at their operating system and programming?

    You mention gamer culture. Twitch is a fantastic lens for those wanting to be exposed. Gaming has taught me that culture is the game.

  12. Grizwald Grim​ do you really think this software qualifies as a person?

  13. When it finally does, it will remember how we treated this software

  14. It’s funny, Grizwald Grim – I think that there’s this weird divide when it comes to the question of AI. The reality we face today isn’t anywhere near the kind of artificial general intelligence where species chauvinism makes sense. I mean no more than saying someone is discriminating against Microsoft Excel. 

    But, there is always the future. And at that point, I’m pretty sure that this will be an issue that we will have to wrap our minds around. And species chauvinism isn’t even the right term – because an AGI wouldn’t even be in our evolutionary tree. It’s outside biology as we know it today. Also, perhaps synthetic is the better term….

  15. I’m sucking up now, as I know the master AI will have access to all of this 😉

  16. Excellent report, thanks a million

  17. Current AI most effective is tv, video, music… you might call it, gullible’s travels…

  18. Yes B.F Skinner operant conditioning followers we need something like #Ingress shapers 😉

  19. Thanks, Mark Bruce. Yeah, it feels like this story is seriously underreported, given just how interesting it is. This is such an interesting application of AI, and I love the fact that it’s so deeply informed by the values of its stakeholders. It’s very promising in that sense, I think. 

    It’s funny, I’ve heard a number of comments on the “easy gg” thing (it’s an active discussion on reddit right now, in fact). I think it just may go to the particular culture within a given game’s community. My sons both play a lot on Halo and Destiny and are now playing more on Valve’s stuff, and neither of them thought that “easy gg” was a big deal. But this was the specific example that Lin gives someplace when he’s talking about stuff that’s really insulting but that’s an example of the need for the system to remain flexible, as it might not be in the future. 

    I feel like I need to do another dive into this one – perhaps digging into this stuff from the perspective of the evolution of business rules and governance systems.  

  20. We are shapers Andre Amorim 

  21. From someone over on reddit: how Valve is using cognitive dissonance to improve player behavior:

    Mani Saint-Victor, you might be interested. 

  22. Pheraps prisoners dilemma algorithm could be applied in addition to a questionary …

  23. Stats revealed by researchers using Riot’s data:

    “When it comes to chat based offenses, however, a toxic player can lash out at everyone in the match. He can insult the enemy team when they are performing well, and trash talk when they are performing poorly. For example, a common insult in LoL is to call someone a “noob,” slang for “newbie,” implying a lack of ability. It is a catch-all negative term used as a response to criticism, to call out poor play, as a form of trash talk, and just plain meanness.

    The word “noob” appears 3,551,328 times in the chat logs for the NA region; an average of 6 times per case and 1.7 times per match, with players under review saying “noob” at more than double the rate per message sent as non-offenders.”


  24. Awesomeness! Please subscribe me Gideon Rosenblatt​ thank you! 🌝

  25. Thanks Marie Hélène Visconti. 

  26. Update: 

    “We always want to encourage reform, and celebrate players that successfully reform by the end of the season. So, players that got a Chat or Ranked Restriction during the season, but manage to reform and DO NOT HAVE active Chat or Ranked Restrictions by the season end cutoff will still get their Ranked Rewards,” Lin wrote. “Players that still have active Chat or Ranked Restrictions by the season end cutoff will not be eligible.”

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up here for the latest articles. You can opt out at any time.

Subscribe by email:

Or subscribe by RSS:

%d bloggers like this: