Making data talk to us

Photo by fauxels from Pexels

I am passionate about communication 📣. I believe that efficient communication is relevant, interesting and has great value. One of the problems today, is that most of the time we are bombarded with messaging that could carry great value for us, but is neither relevant nor interesting.

As machines continue to get smarter, and as we collect more and more data, messaging will become increasingly relevant. The timeliness, placement and value of the message aligns with our current intent, yet we still disregard them, because the way the it is delivered is simply not interesting.

Our biggest challenge is getting data to actually communicate. Being timely and relevant is not enough. We need to make data communicate.

I will try to demonstrate it with my last attempt at learning french.

So, I went to my favorite search engine and typed in “learning French”.

I have indicated that it is now a good time to receive some information about it. I have intent, I’m attentive, I’m ready.

Today, the first result I get out of the 2,370,000,000 is an ad, for expert tutors. While it might be relevant, timely, and of value to me, the text is not interesting, and I move on. I scroll around, click on a few things, get frustrated, and get back to coding…

But what if the result was a video of a tutor most relevant to me, teaching me how to say: “my name is Leo” ? 🦁

Would that have caught my attention?

What if, after I clicked through, I found myself in an interactive language learning course. Friction free, immersive, interesting.

There are a few reasons people wouldn’t think this possible: the first is that video production is hard work (especially when featuring real human subjects); and once it is produced, the video is what is known as “backed,” meaning that it is immutable. The video is not flexible and cannot be adapted for different users.

The next is the sheer number of videos that would be required to deliver such a personalized and special-feeling experience. Think of all the names of people that have entered this search query, and multiply that by the number of tutors… that’s the number of videos…

Making this amount of video would be a full time job in itself for the tutoring school 😁 leaving no time for actual teaching.

All the data required to trigger this video result exists in the very moment that I enter my text query in the search box. This is the trigger required to produce short video responses, in all their possibilities. But then if the search engine doesn’t return the exact result I’m looking for in a split second, I assume it’s broken and go and distract myself on social media. 😎

Dimensionality expansion is what’s missing here. The process of taking data from a lower dimension (Text) and expanding it to a higher dimension (Video). This is exactly what we do at hourone.ai

We make video from text.

We are leveraging the latest advancements in artificial intelligence: deep learning, and GANs. With our system you can give us text, either type it in through the user interface or create it programmatically through our API and immediately, auto-magically 🎩 get an instant video result. Like the video I showed earlier.

We can create all these videos today just by connecting all the data points: The intent, my name, the tutor most suited to me, and a bit of French 👨‍🎨

Hyper Text to Hyper Video is what I believe will make communication in the near future much more interesting, and therefore, effective.

And this is what we are building… amongst other things.