HCAI
Strategist
AI Ethicist
Public speaker & advocate for responsible AI development.
HELLO WORLD
I am Di Le, and I am deeply immersed in human-centered AI design. I bring my strategic insights to the forefront of responsible AI development at ServiceNow and a spectrum of international organizations.
My journey is enriched by extensive experience in human-machine interaction (HCI), where I've led the charge in research and design efforts to democratize autonomous collaborative robots for non-experts.
I've played a pivotal role in crafting the standard for human-AI collaborative workflows, impacting the deployment of autonomous mobile robots worldwide.
Beyond my work in HCAI and HCI, I’ve helped organizations create frameworks and tools that underpin responsible AI practices.
As a technology innovator, an ambassador for Google's Women Techmakers, and a judge for the CES Innovations Awards, I am committed to fostering conversations about responsible AI. I aspire to develop technology that addresses pressing human challenges and embodies the principles of beneficence and benevolence, shaping a future where innovation and ethics converge.
CONTACT
hellol@dibot.io
Linkedin
MEDIA
UXC - HEC University Montreal
From jackpots to AI algorithms: The role of
human-centered design in navigating unpredictability.
This White Paper discusses the human impact and transformative potential of generative AI and ChatGPT, likening the increased accessibility of AI tools to breaking "the fourth wall" between creators and audiences. It explores how this technology is dismantling traditional barriers and prompting new inquiries.
White Paper: The Democratization of AI
White Paper: AI Post-ChatGpt A Progress Report & Future Forcast
This white paper analyzes the lessons learned and the insights gained that can guide us toward a future where we mitigate the risks while maximizing the opportunities for innovation, equity, and shared understanding.
Azimuth: Systemic Error Analysis for Text Classification
Presented at EMNLP.
This paper introduces Azimuth, an open-source, user-friendly tool designed for error analysis in text classification. Highlighting the relative underdevelopment of error analysis tools compared to other machine learning development stages, the paper underscores its importance for creating reliable AI systems. Azimuth offers a systematic approach to error analysis, combining dataset examination and model quality evaluation. It integrates various machine learning techniques, including saliency maps, similarity, uncertainty, and behavioral analyses, to help AI practitioners identify and address areas where models lack generalization.
UPCOMING EVENTS
3/6/25
Halıcıoğlu Data Science Institute
3234 Matthews Ln, La Jolla, CA 92093
3/29/25
TEDx Talk
Details coming soon.