Y Magazine To All: How AI is Helping Share Gospel Messages in Other Languages
Apr 2, 2026
Steve D. Richardson, a computer science professor and machine-translation researcher who led IBM and Microsoft projects before joining BYU, discusses using AI to bring gospel materials to low-resource languages. He talks about recording speech in Xhosa, Twi, and Zulu to train models. He describes scaling PathSAI, dubbing content, expanding globally, and including sign languages to widen access.
AI Snips
Chapters
Transcript
Episode notes
Pathway Students Preserving Languages By Recording Gospel Texts
- Thomas recorded church materials in his native Thuy through the PathSAI project to preserve and practice his language.
- Pathway students across South Africa, Nigeria, and Kenya have recorded over 1,100 hours in 18 languages to train speech AI models.
Decades Of Translation Work Led To Today's AI Breakthroughs
- Steve D. Richardson's lifelong work in machine translation traces back to a 1975 prophetic call and decades of projects at IBM, Microsoft, and BYU.
- Early translation efforts used rule-based systems and massive donated hardware, showing incremental progress until the internet and large datasets accelerated breakthroughs.
Compute Internet And Data Converged To Unlock Large Language Models
- The compute, the internet, and massive multilingual datasets converged to make large language models possible after decades of slow progress.
- Richardson frames this technological timing as providential, saying the explosion in capability arrived on "the Lord's timing."