Creating More Natural, Human-Like Conversations With Google’s LaMDA
Although chat and voice technology has come a long way, conversational AI doesn’t yet have the skills to mimic the free flowing, conversational capabilities of humans. Recent advances have positioned assistants to handle the natural rhythms of speaking. For example, Alexa Conversations allows Alexa to answer several questions at once, remember information already provided, and respond to corrections. Google Duplex’s AI leverages pauses and “ums” to sound more natural to the person on the other end of the call. However, these interactions are usually confined to only one topic or goal. With Google’s ongoing development of its LaMDA software, conversational AI is poised to be more flexible and navigate seamlessly from topic to topic, emulating how people interact with each other.
Header image source
Ask Me Anything
Google has announced that it is developing LaMDA (Language Model for Dialogue Applications), its advanced conversational AI system. LaMDA can converse more naturally with users via chat, such as talking about nearly any topic without prior training. For example, the demo showed LaMDA pretending that it was the planet Pluto and even a paper airplane, answering questions related to these topics. The platform’s text-reading and word prediction capabilities allow it to engage in open-ended conversation that doesn’t follow a predetermined path, while also providing context-relevant answers. This new system aims to make conversation more human-like, paving the way for how voice tech and assistants can be improved. More on PCMag
Snapchat recently unveiled its fourth-generation Snapchat Spectacles, its smart glasses that record videos for the mobile app. However, the new lenses use 3D waveguide displays that provide AR capabilities, allowing users to layer digital objects (via the Snapchat Lens filters) on top of their physical environment. Similar to the mobile app, users can use the wake word “Hey Snapchat” to choose filters and settings. Companies like Facebook are already reimagining how smart glasses might leverage hand and finger gestures for an immersive experience, but Snapchat’s voice feature sets the stage for more voice-enabled AR and VR experiences for everyday social media users. More on CNBC
Voice tech car company Cerence partnered with tooz technologies to bring its voice assistant to tooz’s smart glasses. The glasses include a small virtual screen, positioned over the retina, that can be controlled via voice to show the driver information about traffic, speed limits, and more. Although voice has primarily been positioned for automobile drivers, its utility for those riding motorcycles is compelling — it can promote safer, hands-free retrieval of information, especially when it is paired with a visual interface. Smart glasses are becoming more popular (Snapchat’s AR glasses and the new Echo Frames sunglasses arrived last week), but now companies are framing smart glasses to go beyond leisure and entertainment use. More on Voicebot.ai
Emerging Tech Stories
- Project Starline. Google’s upcoming video chat platform creates a 3D version of the person you are talking to for a more realistic experience.
- Try On. Snapchat’s new app update lets customers use voice commands and gestures to virtually try on clothing.
- Rap Battle. The website uberduck.ai’s technology allows users to generate audio clips of artists including Kanye West and Eminem by typing in any text.
By The Numbers
67% increase in chatbot usage between 2018 and 2020.
(data from Salesforce via VentureBeat)
29.1% of marketers believe voice assistants are extremely important as a marketing channel. (data via Voicebot.ai)
Creating a Voice-First Learning Experience With ABCmouse on Google Assistant
Learning is still fun, even when children are stuck at home! RAIN partnered with Google and Age of Learning to bring Age of Learning’s award-winning ABCmouse experience to Google Assistant across mobile devices, smart displays, and smart speakers. With the custom Action, children embark on a voice-driven journey to master their grasp of the alphabet with letter songs and stories, sing-alongs, and letter tracing activities.