Benchmarking Beyond Borders:
Making AI Testing Truly Global
Current evaluation approaches used by AI model developers often fall
short when tasked with assessing and improving model performance in non-English languages and contexts. Researchers around the world are developing community-led benchmarks and evaluation tools which prioritize participation by native language speakers and
context experts to more effectively evaluate the usefulness and safety of major AI models, yet many of these efforts are happening in silos.
In the lead up to the India
AI Impact Summit in February 2026, the Center
for Democracy & Technology, Cornell
Global AI Initiative, and Masakhane’s
African Languages Hub will host an officially recognized pre-event to the Summit to bring together researchers, policy experts, model developers, and other stakeholders to exchange insights on making evaluations more multilingual and surface avenues
through which this work can occur. Some of the barriers impeding more robust multilingual evaluations — data cost and availability, for example — exist across languages, so exchanging lessons and tactics can help advance evaluation tooling in multiple languages
and contexts. The ultimate goal: to make evaluation more representative of non-English languages and contexts and make tools readily available to model developers and deployers.
This official pre-summit convening will begin with virtual presentations by leaders working on advancing multilingual evaluation efforts around the world. After the presentations, a panel discussion will dive into how these efforts should be supported and what
gaps exist to making these evaluation tools available to model developers and deployers, ranging from industry to public actors.
When: Thursday,
January 29, 2026 from 9:00 - 10:30 am ET
Speakers:
-
Miranda
Bogen, Center for Democracy & Technology
-
Chenai
Chair,
Masakhane African Languages Hub
-
Alexandra
Givens, Center for Democracy & Technology
-
Bennett
Hillenbrand, MLCommons
-
Sara
Hooker, Adaptation Labs
-
Faisal
Lalani, Collective Intelligence Project
-
Chinasa
T. Okolo, Technecultura
-
Roya
Pakzad, Mozilla Foundation
-
Sunayana
Sitaram, Microsoft Research
-
Dhanaraj
Thakur, George Washington University
-
Aditya
Vashistha, Cornell University
Accessibility:
If you have questions, concerns, or access needs, please contact
eve...@cdt.org in advance of the event.
|
|
|
|
|
|
|
|
