Promoting Prosociality with AI

Recommends moderation strategies in real-time that are context-specific, theory-backed, validated, and explainable.

Reward prosociality

HUGO handles both hate speech and prosocial behavior

Suggest responses

HUGO generates context-specific and explainable responses

Configure behavior

You can customize response styles, languages, strategies and more

Promoting Prosociality

Mitigating Antisociality

Multilingual Listening

About HUGO

Unlike any current moderation tool, HUGO recognizes opportunities to intervene and offers suggestions for how to intervene.

Understanding the User's Comment

HUGO detects not only toxic content, but also prosocial content and more nuanced social regard at the word and phrase level.

Responding with Empathy

HUGO produces natural responses, in any language, to flagged comments (both anti- and pro-social) warning or rewarding the comment's author, while leveraging social science-theoretic notions of empathy, interdependency, productive discourse and the like. We uniquely computationalize extant social science theories to generate natural language responses.

Evaluating Prosociality

HUGO combines various measures of politeness, toxicity and social regard to generate a unified metric of "civility" that captures social and moral expectations in discourse.

Measuring Civility

HUGO computes a novel theory-based measure of civility for both individual comments as well as for a group of comments

Private Civility

Linked with interpersonal politeness. Violations include name-calling, personal insults, foul language, and targeted toxicity

Public Civility

Linked with a collective sense of politeness. Violations involve behaviors that threaten democratic values, deny people personal freedoms, stereotype social groups, and elicit polarization

Green and Red Measures

Green represents the total amount of evidence from a set of social sensors supporting civility. Red represents the total evidence supporting incivility

We are computational social scientists

We love a debate, but prefer a healthy one. We love a good vent, but prefer a respectful one. We love a connection, but prefer a prosocial one

Sonja Schmer-Galunder

Social Anthropologist and Principal Researcher at SIFT, specializing in sociocultural applications and implications of AI

Vasanth Sarathy

Computer/Cognitive Scientist and Senior Researcher at SIFT, specializing in Generative AI with social, creative, and ethical reasoning capabilities

Laurel Bobrow

Cognitive Scientist and Researcher at SIFT, specializing in developing data-driven approaches to AI

Rick Freedman

Human-AI interaction Researcher at SIFT, specializing in AI algorithms for planning, activity and intent recognition

HUGO stands for Humanitarian User Group Oversight. The HUGO name and logo were inspired by Victor Hugo's "Hunchback of Notre Dame" and his cry for sanctuary. The image mimics the center of the Notre Dame rose window with colors coming from Inglehart-Welzel cultural map of the world

Made with