Skip to main content

The Future of Machine Translation

What does a future with AI translation look like?
Photo by Helen Kayda

The difference is almost indistinguishable. Attendees scratch their heads, struggling to distinguish the legitimate President Nelson talk from the talk that was translated into Spanish and back into English with Google Translate. Some answer correctly, but most are stumped, proving the increasingly impressive skills of artificial intelligence (AI).

Professor Steve Richardson (Machine Translation) used this comparison exercise to show just how advanced AI is becoming. Richardson presented “Machine Translation: Hype vs. Reality” to a crowd of humanities and technology faculty (and enthusiasts) on February 16 at the Humanities Center Colloquium. He explained just how far AI translation has come in recent years—and how it is revolutionizing the field of translation, especially within the Church.

History of Machine Translation

Machine translation (MT) has evolved extensively in the last 75 years. From the ’50s to the ’80s, rule-based technology was utilized to distinguish words with different meanings depending on the context.

Now, the focus lies on neural translation, which predicts the likelihood of a sequence of words and is approaching human-level accuracy. Companies like Google, Microsoft, Amazon, and Meta have all been on the front lines of the MT movement, using technologies like Google Translate, ChatGPT, and Wav2Lip.

Application in the Church

As the manager of machine translation for 10 years at The Church of Jesus Christ of Latter-day Saints, Richardson highlighted how the Church uses MT to gather Israel and spread the gospel to all ends of the earth.

With AI, the Church hopes to translate content between all languages without English being the lingua franca (or common language among people of diverse speech), whether that be for general conference, the Book of Mormon, or other forms of Church literature. “What would it be like if there were no language barrier, so somebody in China could share the gospel with somebody in Brazil and not have to worry about the fact that they have to both speak English?” he asked.

Additionally, the Church wants to use this technology for MTC missionaries who don’t speak English but want to attend devotionals. So how does it work? Richardson played a video of Elder and Sister Renlund giving a devotional. The AI translated the speech in real time. “It may not be as high quality, but if you can match this against an existing script that they’re following, which all the Church leaders do, then wow, we can get much better machine translation,” he said.

Synchronized Speech

AI can also make speech sound human (not robotic) by merging audio files with videos. Richardson noted that he showed the technology to Church leaders, and “they thought it was humans who were talking.” AI’s capacities constantly evolve and improve, shaping the possibilities of the future.

Synchronization also proves a useful tool for the future, allowing for technology similar to deepfakes, a digitally altered video of a person where they appear to be someone else. Richardson showed video example of a girl speaking English; however, with the Wav2Lip technology, it translated her words into Japanese and made her look like she was speaking it. For the Church, this possibility means that international audiences could see apostles, prophets, and other Church leaders sounding and looking like they speak their native language.

The Future of AI

AI is the way of the future, but Richardson assures that it will prove a great resource for the Church and the world. “You can be afraid of it, or you can just embrace it and say, ‘Okay, it’s going to need some human help’” he said.

The technology continues to improve, making it easier for translators to use and becoming a valuable asset. “BYU is uniquely positioned to contribute to this work, impacting scholarship and benefiting the Church and society,” Richardson said. As AI develops, one thing remains certain: artificial intelligence is the way of the future.