2020 Voice Assistant roundup

At the end of 2020 our CEO, Charlie Cadbury, was interviewed by Erik J Martin looking back at an important year in the rise of voice assistants

1. Overall, how would you summarise and describe the state of artificial intelligence impacting speech technology to date?

Between 2018 and 2020 we saw an increase of UK adults having access to a voice assistant at home rise from 14% to 34% (Audience Project) and we now see research analysts like Juniper Research saying 2020 will see $20bn of voice transactions moving to $164bn by 2025.  This is the reason everyone’s ears are pricking up right now…

2. What were some of the key highlights in 2020 when it comes to the state of AI in speech technology? Please nominate a few of the most important.

developments/announcements/innovations in this arena (related to machine learning, speech-based systems run by AI, etc.)

In the UK at least, May was a watershed moment.  The way to complete voice transactions changed.  Previously one had to allow transactions by confirming using the mobile app.  From May 2020 that could be done using voice permissions.  We (at http://sayitnow.ai)  used that and a new way to target customers listening on a smart speaker to drive charitable donations.  We received a lot of positive press about this approach.

3. What is the state of AI adoption, and why has adoption been challenging up to this point?

As we mentioned a lot of homes have a smart speaker but what are they boeing used for?  88% are just being used for streaming music (voicebot.ai).  However, during COVID these Voice Assistant systems have been trusted more and more as a source of truth and 50% of 18-34 year olds say they are using their smart speaker more to find out about about news and information than before the pandemic (Edison Research)

4. What is the speech and voice recognition market expected to be worth in the years ahead – any stats/links you can provide?

Our killer stat here is that voice commerce will move from $20bn in 2020 to $164bn by 2025 (Juniper Research)

5. What challenges and shortcomings need to be addressed in 2021 and beyond when it comes to AI and speech technology? What other challenges are related to AI and speech technology (e.g., businesses haven’t yet figured out how to use AI in a targeted way to solve business problems, etc.)?

Discoverability is the biggest challenge and we covered that in depth in our blog “Voice, from crap wizard to personalised genie

6. Looking ahead, what do you predict when it comes to AI and speech technology and 2021 and beyond? What progress/innovations might we see in the coming months/years and why?

We will see Siri opening up its development environment so that third party developers like us can create voice first experiences on that platform.  That will be transformational as the Apple PR system is world class.  The technology is already here.  NFSI (name free skill interaction) is the other shift that will allow third party voice experiences to be served without the user having to ‘remember’ the words to say to get there.

7. What your prediction: Will we begin to see better integration with other technologies, like facial recognition and voice recognition?

We need to pay attention to the ‘uncanny valley’ and do we really want devices in our homes recognising our faces (BTW our phones do it all the time).  There is a lot of benefit to deliver to consumers and if done correctly can unlock lots of tasks for us all that we currently spend on mundane tasks.  The clue is in the name ‘assistant’’

8. To what extent will AI both enable and destroy deepfakes in the months and years ahead – what your hunch here?

Deepfakes are here to stay and a big problem.  The cure in my opinion is in self regulation of platforms (which will do so to earn much needed trust) and consumers who are increasingly savvy.

9. What can we expect regarding AI regulation (internationally, federally, etc.) in the months and years ahead?

Much change, education and understanding.  There is not, should there not, be a clear path.  This needs to be an iterative process where all parties learn how to learn, adapt and change together.

10. How and to what extent will AI increasingly be applied to automatic speech recognition looking ahead?

This feedback loop is imperative to the rapid improvement of a cloud based system such as voice.  Every day there are billions of lessons learned nd applied to make these systems better for tomorrow.  Don’t underestimate that is the now, not the future.

11. Any other tips, thoughts, comments, or ideas on this topic you’d like to add?

Stay in touch. We won UK & EU ‘Alexa Cup’ and are leading the conversation in this space.  We love a conversation so please ask any other questions you may have.

About Say It Now

With a heritage of enterprise software delivery experience, Say It Now specialises in understanding natural language so that computers can talk to people in the same way that people converse with each other.  This helps brands and consumers to benefit from automated communications, which are becoming increasingly relevant as more people have smart speakers in their homes and are becoming comfortable and confident using them.

Launched in 2018, the company provides consulting and software development services to enable transactions and insights using voice assistants on smart speakers (Amazon Alexa and Google Home) and chatbots on messaging platforms such as Facebook Messenger, WhatsApp and SMS.

The company won the UK and European rounds of Amazon’s Alexa Cup competition, and achieved the bronze medal in the final held at the end of July 2019.