The "Aha" Moment of Real-Time Language Translation

Yesterday I attended a high school graduation ceremony for Comp Sci High, a relatively new charter school in the South Bronx. With only its third-ever graduating class of approximately 100 seniors, it was a big day for the entire school community. Many students are matriculating to college as the first in their family to do so.

Like a lot of communities in their neighborhood, the school has a large Spanish-speaking population. They have always done their best to accommodate families who are unable to understand English, often with a human translator at large events.

But at yesterday's graduation ceremony they did something a little different. Before the official programming began, a staff member went on stage and asked (in Spanish) how many people in the audience are native Spanish speakers. About a third of people raised their hands.

Then they projected a giant QR code onto a screen that linked to a real-time, AI-powered translator bot.

"Scan this QR code with your photo," they instructed. "Click the link, and you'll be able to see everything that we are saying upfront with real-time captions en español."

Here's what it looked like in real time when the first speaker arrived.

The English text -- which was being pulled in live from a Zoom meeting transcript they were generating live -- would appear on top.

Then the Spanish captions would appear within seconds right below.

It was the coolest thing I've seen in live translation since those polarized, auto-caption subtitles built into the back of every seat at the Metropolitan Opera.

This is what it looks like to meet your community where they are with AI.

In a city like New York, with 3 million+ foreign-born residents, as recently as 2021, 42% of students in enrolled in New York City Public Schools have a primary home language other than English. The city is already all-too-familiar with the problem of disseminating information about city resources to people who don't speak English.

But it does get you thinking: What else is possible?

Last night when I got home from the graduation, I had an email in my inbox from NYC Public Schools informing me about kindergarten registration information sessions. I have a four-year-old at home, which means that it's time to start thinking about my options for elementary school.

They offered three information sessions for parents to learn about our options: One in Spanish, one in Arabic, and one in English, with a variety of other translating options provided live.

Unfortunately, my husband and I will be out of town on August 22, when they are hosting their English speaking session.

So I wonder -- could I attend the Spanish-speaking session and live translate it for myself with AI? Could anyone attend any session and translate it to meet their needs? Rather than host three sessions, could they just host one?

I'm going to try it out.

While I'm relatively new to embedding myself in the AI ecosystem, one of the things I'm quickly noticing is how much the idea of coming up with practical applications is becoming a creativity problem, as opposed to a technological problem. (This, by the way, is almost the opposite feeling from my work in web3 -- where it seems creativity has no bounds, but the technology needs to catch up to meet the needs of many idealized use cases.)

This is a very fun problem to have if you like solving problems but you never learned how to code. (ie: me) Even if you're not a software developer, AI lets you solve problems with technology. You do not need a degree in computer science to come up with the idea to record a presentation with Zoom, create a transcript, then run it through a translator bot.

But you do need to be constantly asking the question, "How might technology solve this problem?" And then be willing to give it a shot to see for yourself.