Don’t Try This at Home: Honest Tales of AI Failures

Track: Multilingual AI | AI3 |   Everyone |
Wednesday, June 4, 2025, 3:00pm – 3:45pm
Held in: High Live 1 (Plenary)
Presenters:
Belén Agulló García - Terra Localizations 
Miriam Moyano Reifenrath - Riot Games 
Marta Nieto Cayuela - Uber
Moderator: Marina Pantcheva

Most conference presentations highlight success stories of AI-powered solutions. However, behind every success lies a series of setbacks. This panel focuses on these often undisclosed failures. A team of experts will share their honest stories of AI experimentations that didn’t go as planned. They will explore these failures from content owner and service provider perspectives across various localization domains—from multimedia to text. The discussion will cover why AI failed and how humans can help overcome the challenges. Join us to learn how AI failures can drive innovation and mitigate risks connected to operationalizing AI.

Key Takeaways:

Attendees will discover that many companies are facing and overcoming similar challenges and that collaboration and shared knowledge are crucial for achieving collective success; exploring why and when AI fails sheds light on the importance of human involvement and the need for diverse, high-quality data. Attendees will learn how these factors work together to improve AI’s performance in localization; the panel will help attendees manage the expectations of their stakeholders who might overtrust the potential of Gen AI and the immediate impact that it can have on operations.