Spirit AI: Fighting Bullying and Humanizing Characters in The Gaming Industry

Spirit AI uses NLP to help gaming companies create inclusive online communities and humanize their digital characters

If you are a gamer, you must have interacted with a character at some point while playing. Often, you will realize that these characters are very computerized, responding in a weird inhumane way, not reacting to the environment, just programmed to say few things regardless of what you do in the game or your progress or how you got to the level you are at in the game. If you are also a gamer, you must have been part of an online community where people share thoughts and strategies and just connect outside of the virtual world (particularly for e-games for example). Unfortunately, often you find that these online social communities become a space to abuse and bully opponents.

Spirit AI was founded to try and solve these issues. Spirit AI is an AI-based platform based out of London, founded in 2016, offering 2 main services: Ally and Character Engine. In a nutshell, Ally is an online community management tool that can understand people’s words and context and identify bad behavior/ award good behavior. On the other hand, Character Engine is about creating characters that understand the story and react according to the progress of the player, in a human-like way.

AI Technology Modules:

In order to enable these 2 services, Spirit AI uses 3 main technological modules that allow for AI capabilities:

  1. Natural Language Understanding: NLU is at the core of the companies AI capabilities. It is needed to understand people, their motivations, and nuances in social interactions and build sophisticated behavioral classifiers that allow multiple layers of contextual sensitivity understanding
  2. Natural Language Generation: Generating natural language is another key capability that Is required for the characters in the game to interact in a human-like manner with players. This module is key as it enriches the human-computer interaction, something that could be useful for other applications/ industries
  3. Knowledge Management: flexible and dynamic management of the AI knowledge environment is key is quickly adapting to new applications/ new clients/ new services and to changing rules and environments

Ally: Creating Safer and More Inclusive Online Communities

Ally is an online community management tool that utilizes different technologies to identify bad behavior, award good one, and understand how the online community related to a game is operating. Ally has multiple features that create tremendous value for clients:

  1. Context-Driven Analysis: while analyzing voice and text that is exchanged between players/ characters, Ally analyzes different contextual layers such as the relationship between characters, phrases, messages, and actors. The contextual analysis allows Ally to understand how people communicate, detect specific behaviors through NLU. Ally can identify aggression, harassment, bullying and can focus on hate speech, white nationalism, grooming, fraud, and other targeted speech
  2. Moderation Workflow: Ally has a built-in moderation workflow that is robust, efficient, and user-friendly.
  3. Powerful Automation Tools: Ally is highly automated. The clients can easily configure and customize the rules that Ally follows in identifying bad behavior
  4. Deep Insights Into Community Trends: Ally also provides insights into community trends, mainly related to messages, stories, actors, actions, moderators, automation, and behavior

Ally offers a lot of benefits through these features and services: Ally is highly customizable to the customer’s need, uses cutting-edge NLU technology that is regularly updated, rapidly adjusts to changing trends and behaviors, and uses powerful searching tools to scan through millions of messages. Through Ally, customers can eliminate toxic messages and nurture an inclusive community.

Character Engine: Humanizing Characters

A character engine is a toolset that builds characters with a certain level of intelligence. These characters improvise and adapt based on the narrative requirements and the player’s input. These characters will show emotions and communicate naturally with players through NLP (using both modules above NLU and NLG).

Challenges and Opportunities:

Spirit AI faces some challenges as it grows its services:

  1. Localization: while expanding into new geographies, a big challenge that Spirit AI faces is building a clear understanding of local culture and social norms. In some countries, some language can be perceived as aggressive while in others it is not. Or certain words that are can be attributed to bad behavior only exist in specific languages/ cultures. To solve these issues, Spirit AI combines its AI automation with human capabilities that will enable the AI to unlock these cultural nuances
  2. Profitability: through research, it was hard to understand whether Spirit AI has reached profitability or proven its business model. However, like any company tapping into new realms of technology, my hypothesis is that the company did not yet recover all the investments it made in its product. This will only happen as the company broadens its client base

On the other hand, Spirit AI can tap into a multitude of new opportunities:

  1. New Industries: Spirit AI can expand its capabilities beyond the gaming industry. Practically, any industry that has an online community aspect can actually utilize Ally (e.g. Facebook or even professional services companies that have online communities)
  2. New Products: The different modules that the company built to enable its AI capabilities can be sold separately to any company in any industry that might require NLU/ NLG

While Spirit AI is still relatively nascent, the potential it has is something to look out for. The values and objectives of the company are a first step towards building more inclusive online communities, something that big players such as Facebook and Instagram have failed to do.

Sources:

https://builtin.com/artificial-intelligence/ai-games

https://www.spiritai.com/

https://www.bestpractice.ai/studies/spirit_ai_introduces_a_character_engine_to_create_characters_that_enable_players_to_express_themselves_better_using_natural_language_processing#

https://www.crunchbase.com/organization/spirit-ai

https://venturebeat.com/2018/06/17/how-spirit-ai-uses-artificial-intelligence-to-level-up-game-communities/

https://www.pocketgamer.biz/news/68264/spirit-ai-reveals-slew-of-updates-to-social-behaviour-tool-ally/

Previous:

Netflix: Optimizing the user experience and investment decisions using Machine Learning

Next:

AI as a safeguard against cyberbullying

2 thoughts on “Spirit AI: Fighting Bullying and Humanizing Characters in The Gaming Industry

  1. This is very interesting, thank you for sharing Karl! On your point of profitability, do you know if Spirit AI has developed partnership with big game developers (i.e. EA, Nintendo, etc.)? Have they acquired customers outside the gaming community? The technology Spirit AI is developing could be attractive to online communities outside gaming, such as social media platforms and online marketing communities that are often infiltrated by bullies and other problematic characters.

  2. Thanks Karl! Very interesting read. I agree with Juan Carlos, there are probably other attractive use cases such as social media – my only concern is that it (as with most technologies) could be abused. I’m guessing that in the wrong hands this tool could be used for censoring purposes, for example using AI to identify anti-government speech or protest planning. Hopefully Spirit AI has this potential use case in its radar and is taking steps to avoid it.

Leave a comment