Fei-Fei Li, chief scientist for Google AI, told conference attendees that Google is committed to making it simpler to embed AI into business processes and applications using, for example, a new Contact Center AI service as well as additions to Cloud AutoML, a service that makes reusable AI models that can be customized available to developers. Those additions make vision, natural language and language translation services invokable via an application programming interface (API).
Google also announced that a third generation of custom processors, known as Tensor Processor Units (TPU), is now available in alpha. Li says the second generation of those processors, announced last year, is now generally available on the Google Cloud Platform (GCP). Those processors are optimized to run AI models built using the open source TensorFlow framework developed by Google.
Google also unveiled Edge TPU, a new processor that will be employed in a pod to extend AI apps to gateways and connected devices.https://o1.qnsr.com/log/p.gif?;n=203;c=204663295;s=11915;x=7936;f=201904081034270;u=j;z=TIMESTAMP;a=20410779;e=i
The Google Contact Center AI service, available in alpha, has already drawn support from Twilio, Cisco, Mitel, Vonage, RingCentral and others that have pledged to embed AI cloud service within their contact management platforms. To facilitate the development of those applications, Google also announced this week an update to Dialogflow Enterprise Edition designed to make it easier to infuse virtual agents enhanced by AI models into customer management processes and applications.
Google also moved to make it simpler to build and deploy machine learning models based on massive, structured or semi-structured datasets directly inside BigQuery using simple SQL statements.
Li says AI will soon be infused into just about every business process imaginable.
“AI is transforming industries all over the world,” says Li. “It’s our goal to democratize it.”
Naturally, the level of AI sophistication within most companies varies widely. But as it becomes increasingly possible to invoke an AI model via an API using a few lines of code, applications that don’t have AI capabilities will soon be considered largely obsolete.