27/02/2020
Google: Meet Meena, the Lifelike ChatBot.
Jupiter Research has projected that there will be 8 billion digital voice assistants in use by 2023, this is up from 2.5 billion at the end of 2018. Advances from Amazon, Apple, Google and Samsung will drive this across numerous surfaces from smartphones to smart devices (TV's & speakers) to wearables.
Current voice assistants still have a hard time interpreting what it is that consumers ask them to do and their accuracy rate has declined year over year according to Perficient Digital. Recognizing this, a team at Google Research published a paper last week that might be the start to solving to this critical issue.
The Next Web summarized the paper by writing, "The conversational neural network with 2.6 billion parameters can chat with people better than any AI generator out there. The team trained the model with 40 billion words — 341 GB of text data including social media conversations — using the seq2seq model.
"Seq2seq is a variation of Google’s Transformer — a neural network that compares words in a paragraph to each other to understand the relationship between them.
"Meena has a single evolved transformer encoder block and 13 evolved transformer decoder blocks. While encoder blocks help it understand the context of the conversation, decoders help it to form a response. Google claims Meena has 1.7x more model capacity and was trained on 8.5x more data than OpenAI’s GTP-2.
"The team of researchers also devised a new matric called Sensibleness and Specificity Average (SSA) to measure how sensible and specific a conversation or a response is."
What does this mean for brands? With the hardware side of conversational technology now easily embeddable in anything at scale due to Amazon's AWS IoT Core advances last year combined with this new approach to make the agent that lives on the device more intelligent, the future of voice interaction looks bright.
As we've been saying for some now, once consumers come to expect that the assistant be there auto-magically, brands will need to understand how to build experiences to meet them in these new, un-tethered moments, as doing so will unlock new contexts and opportunities to provide value.