Using AI to serve LGBTQ youth in crisis: Findings from The Trevor Project

  • Image from iOS.jpg
    Riley Leight
  • Oct 13, 2021

The Trevor Project offers support for LGBTQ youth who might not have anywhere else to turn—and they’re constantly innovating to carry out that mission. Through machine learning and a new AI conversation simulator, they’re training volunteers to engage with every contact in the most effective way possible. And rather than replace their essential human touch with AI, they’re using technology to enhance the counselor training process instead.

Adjust text size

Content warning: This article discusses The Trevor Project’s work to address suicide rates among LGBTQ youth. We encourage discretion for readers.

Over the organization’s 23-year existence, The Trevor Project’s leaders, staff, and volunteers have seen firsthand how communication can save lives. 

In the U.S., suicide is the second leading cause of death among young people, and LGBTQ youth are more than four times more likely to attempt suicide than their peers. Intervention from a supportive adult can be lifesaving for young people experiencing mental health issues, suicidal thoughts, and struggles related to identity; in fact, having at least one accepting adult can reduce an LGBTQ young person’s risk of suicide by 40 percent. That’s why The Trevor Project devotes an incredible amount of time and resources to engage with their community—while simultaneously building new and better ways to do so.

Young people in crisis can reach The Trevor Project via channels like phone, text, and webchat, depending on which is safest and most accessible to them. Earlier this year, we wrote about how The Trevor Project relaunched their communications platform using Twilio APIs like Programmable Messaging—which contributed to a 22 percent increase in the number of callers they were able to serve. CEO and Executive Director Amit Paley appeared at SIGNAL to provide even more context about the impact of this improvement. He explained how, through a series of intake questions gathered via Twilio-built text and web chat, they developed a risk-assessment system to triage incoming messages. The system uses AI and machine learning to sort messages into queues and ensure that those at the most imminent risk of self-harm are able to reach a counselor as soon as possible.

Now, they’re continuing to innovate by using engagement with an artificial intelligence (AI) system to train their counselors on engagement best practices.

The AI simulator innovating crisis communication training

All of The Trevor Project’s work is high-stakes, which means they have to take a careful approach to any technology they roll out—including AI. Human connection is core to what they do. So rather than try to replace it, they apply AI to their training and operations programs to help them scale and improve. 

Their first foray into this type of technology use case was building a chatbot-style tool named Riley, an AI persona representing a young person in crisis used to train counselors. It operates separately from the triage chatbot but relies on similar data patterns and machine learning.

Details about their life, current challenges, and emotions have been used to guide the AI’s communication and provide depth.

Before a counselor is asked to offer guidance to a real person, they can interact with Riley—in addition to other human-led roleplay conversations—as a way to safely cultivate their communication strategies and develop their own style. The AI has been trained to resemble a real interaction as closely as possible, which helps aspiring counselors understand and experience the challenges of crisis communication before they are speaking directly with young people reaching out to Trevor’s services.

After an interaction with Riley, transcripts are reviewed and discussed to understand how to apply the practice conversation to real engagement—offering counselors invaluable experience-based lessons that help them better serve people in crisis.

This tool has allowed The Trevor Project to scale and optimize its counselor training program to serve more youth, furthering their goal of saving an increased number of lives. 

In their SIGNAL 2021 session, The Trevor Project’s technology leaders Dan Fichter and John Callery discussed their organization’s unique, human-centric approach to applying AI to its life-saving work—highlighting key learnings for other organizations.

Discover more with The Current

With the right technology, you can apply scale and efficiency for the greater good in 2022 and beyond. Learn more about the trends and products to help you get started in our latest edition here.

Dive into the rest of our post-SIGNAL edition.

Keep on reading
I want to see more about: 
  • Editions
  • Industry
  • Product
  • Region
  • Solution
  • Use case
Communication for good | Spring 2021
  • Communication for good | Spring 2021
  • COVID-19 and the new normal | Winter 2020
  • Digital trust | Summer 2021
  • Pre-SIGNAL special | Fall 2021
  • Retail in 2021 | Summer 2021
Let's go
Image from iOS.jpg

Riley Leight

Riley is a full-time writer based in Annapolis, Maryland.