Under the Hood: AI at Speexx

Artificial intelligence, like in all industries, is a common buzzword in the language learning market, thrown around often by language companies who promise their users more personalized products. Both B2B and consumer language companies proclaim that, thanks to AI, people can reach fluency by using their apps.

At Speexx, AI is not merely a buzzword: We have long understood that the one-size-fits-all product of the past is not ideal; everyone learns at different speeds, begins learning at different levels, and has unique, individual preferences when it comes to how they’d like to learn (i.e., conversational face-to-face training vs. quick exercises on a mobile device).

Several years ago, we pivoted in the way we developed software; we were replacing one legacy system after the other before we decided to scrap the whole thing and start at the beginning. It was certainly a huge effort, but one major result from going back to basics and building a brand new foundation was that we were able to create a booking system for virtual training, which uses AI to automatically match users and coaches based on language level, job role and title, learning behaviors and preferences, trainer profile, interests and more. Indeed, this new structure really uses AI to benefit Speexx users and customers.

Speexx’ learning feed uses AI to provide personalized learning and spaced repetition for vocabulary training, as well as for writing and listening skills and even pronunciation. For pronunciation training in particular, we developed a proprietary speech recognition software specifically for the purpose of learning a language.

We also use AI to predict user behavior. One example of this is our intelligent system response to potential drop-outs. Based on anonymized data from our eight million professional users, we identify user behavior patterns and then take targeted, customized action to prevent users from dropping out.

We’re excited to see what the future of AI will bring for language learning. We expect that within the next 10 years or so, technology will emerge that lets users learn new languages even more effectively and efficiently – software that accurately mimics conversation, promptly points out and analyzes mistakes, writes on whiteboards, and so on – all things that “real” teachers do.

Additionally, it’s exciting to keep tabs on AR and VR, which we anticipate will have the ability to take us to any place in the world to learn a language in a geographical or role-based context. For instance, if you’re learning French, you could “go to” Paris, or Bordeaux, and sit in a café or vineyard and practice your new skills with the local native speakers.

In the meantime, however, humans need not worry – while tech will have the opportunity to provide improved learning environments, to better facilitate collaboration, an optimal learning atmosphere will still need to be guided by human sense; this will remain in the hands of humans for the next couple of years.

So although we are not quite ready for 100% artificial intelligence language teachers, some in the private consumer space and the enterprise market are already using AI throughout their platforms. It will be interesting to see how this develops as the intelligence of our virtual assistants expands further.

Read our whitepaper on Putting Personalization and Flexible Learning at the Heart of Your Business and learn more about how AI helps create a user-centric learning experience.