US 11,990,124 B2
Language model prediction of API call invocations and verbal responses
William J. Byrne, Mountain View, CA (US); Karthik Krishnamoorthi, San Jose, CA (US); and Saravanan Ganesh, Mountain View, CA (US)
Assigned to Google LLC, Mountain View, CA (US)
Filed by Google LLC, Mountain View, CA (US)
Filed on Dec. 22, 2021, as Appl. No. 17/645,739.
Prior Publication US 2023/0197070 A1, Jun. 22, 2023
Int. Cl. G10L 15/26 (2006.01); G10L 13/02 (2013.01); G10L 15/06 (2013.01); G10L 15/18 (2013.01); G10L 15/197 (2013.01); G10L 15/22 (2006.01); G10L 15/30 (2013.01)
CPC G10L 15/197 (2013.01) [G10L 13/02 (2013.01); G10L 15/063 (2013.01); G10L 15/22 (2013.01)] 28 Claims
OG exemplary drawing
 
1. A computer-implemented method when executed by data processing hardware causes the data processing hardware to perform operations comprising:
obtaining an utterance from a user comprising a user query directed toward a digital assistant;
generating, using a language model, a first prediction string based on the utterance;
determining whether the first prediction string comprises an application programming interface (API) call to invoke a program via an API;
when the first prediction string comprises the API call to invoke the program:
calling, using the API call, the program via the API to retrieve a program result responsive to the utterance;
receiving, via the API, the program result;
updating a conversational context with the program result, the conversational context comprising the utterance; and
generating, using the language model, a second prediction string based on the updated conversational context; and
when the first prediction string does not comprise the API call, providing, for output from the digital assistant, an utterance response to the user based on the first prediction string.