Generating and/or utilizing a machine learning model in response to a search request

Abstract

Implementations relate to providing, in response to a query, machine learning model output that is based on output from a trained machine learning model. The machine learning model output can include a predicted answer to the query, that is predicted based on the trained machine learning model. The machine learning model output can additionally or alternatively include an interactive interface for the trained machine learning model. Some implementations relate to generating a trained machine learning model “on the fly” based on a search query. Some implementations additionally or alternatively relate to storing, in a search index, an association of a machine learning model with a plurality of content items from resource(s) on which the machine learning model was trained.

1 . A method implemented by one or more processors, the method comprising:
 receiving, via a network interface, a search query submitted by a client device of a user in response to user interface input that is provided by the user at the client device;   determining that search resources that are responsive to the search query fail to satisfy one or more criteria;   in response to receiving the search query and in response to determining that search resources that are responsive to the search query fail to satisfy the one or more criteria:
 processing the search query using a machine learning model to generate a predicted answer to the search query; and 
 transmitting, to the client device and in response to receiving the search query, machine learning model output that includes the predicted answer that is generated using the machine learning model,
 wherein transmitting the machine learning model output causes the machine learning model output to be rendered, in an interface at the client device, with an indication that indicates the predicted answer of the machine learning model output is a machine learning model generated prediction. 
 
   
     
     
         2 . The method of  claim 1 , further comprising:
 in response to receiving the search query and in response to determining that the search resources that are responsive to the search query fail to satisfy the one or more criteria:
 causing a prompt to be rendered, at the client device, that informs the user that the machine learning model output will be subsequently provided,
 wherein transmitting, to the client device, the machine learning model output, occurs after causing the prompt to be rendered. 
 
   
     
     
         3 . The method of  claim 2 , further comprising:
 in response to receiving the search query and in response to determining that the search resources that are responsive to the search query fail to satisfy the one or more criteria:
 causing search results, that correspond to the search resources that are responsive to the search query, to be rendered at the client device and along with the prompt, 
 wherein transmitting, to the client device, the machine learning model output, occurs after causing the search results to be rendered. 
   
     
     
         4 . The method of  claim 3 , wherein when the machine learning model output is rendered in the interface, the search results are rendered, in the interface, positionally below rendering of the machine learning model output in the interface. 
     
     
         5 . The method of  claim 3 , further comprising:
 selecting the machine learning model, wherein selecting the machine learning model is from among multiple machine learning models;   wherein processing the search query using the machine learning model to generate the predicted answer to the search query is responsive to selecting the machine learning model.   
     
     
         6 . The method of  claim 1 , further comprising:
 in response to receiving the search query and in response to determining that the search resources that are responsive to the search query fail to satisfy the one or more criteria:
 causing search results, that correspond to the search resources that are responsive to the search query, to be rendered at the client device, 
 wherein transmitting, to the client device, the machine learning model output, occurs after causing the search results to be rendered. 
   
     
     
         7 . The method of  claim 6 , wherein when the machine learning model output is rendered in the interface, the search results are rendered, in the interface, positionally below rendering of the machine learning model output in the interface. 
     
     
         8 . The method of  claim 1 , further comprising:
 selecting the machine learning model, wherein selecting the machine learning model is based on the search query and is from among multiple machine learning models;   wherein processing the search query using the machine learning model to generate the predicted answer to the search query is responsive to selecting the machine learning model.   
     
     
         9 . The method of  claim 8 , wherein the machine learning model output further includes an interactive field, and further comprising:
 receiving, via the interactive field and after rendering of the machine learning model output, additional user interface input that varies from the search query;   in response to receiving the additional user interface input:
 processing the additional user interface input, using the machine learning model, to generate an additional predicted answer to the additional user interface input; and 
 causing the machine learning model output to be updated to include the generated additional predicted answer. 
   
     
     
         10 . The method of  claim 1 , wherein the machine learning model output further includes an interactive field, and further comprising:
 receiving, via the interactive field and after rendering of the machine learning model output, additional user interface input that varies from the search query;   in response to receiving the additional user interface input:
 processing the additional user interface input, using the machine learning model, to generate an additional predicted answer to the additional user interface input; and 
 
 causing the machine learning model output to be updated to include the generated additional predicted answer. 
   
     
     
         11 . The method of  claim 1 , wherein the one or more criteria include one or more quality criteria. 
     
     
         12 . A method implemented by one or more processors, the method comprising:
 receiving, via a network interface, a search query submitted by a client device of a user in response to user interface input that is provided by the user at the client device;   selecting a machine learning model, wherein selecting the machine learning model is based on the search query and is from among multiple machine learning models;   in response to receiving the search query and in response to selecting the machine learning model:
 processing the search query using the machine learning model to generate a predicted answer to the search query; and 
 transmitting, to the client device and in response to receiving the search query, machine learning model output that includes the predicted answer that is generated using the machine learning model,
 wherein transmitting the machine learning model output causes the machine learning model output to be rendered, in an interface at the client device, with an indication that indicates the predicted answer of the machine learning model output is a machine learning model generated prediction. 
 
   
     
     
         13 . The method of  claim 12 , further comprising:
 receiving, via a network interface, an additional search query submitted by the client device in response to additional user interface input that is provided by the user at the client device;   selecting an alternate machine learning model, wherein selecting the alternate machine learning model is based on the additional search query and is from among the multiple machine learning models;   in response to receiving the additional search query and in response to selecting the alternate machine learning model:
 processing the additional search query using the alternate machine learning model to generate an additional predicted answer to the additional search query; and 
 transmitting, to the client device and in response to receiving the additional search query, machine learning model output that includes the additional predicted answer that is generated using the alternate machine learning model,
 wherein transmitting the additional machine learning model output causes the additional machine learning model output to be rendered with an indication that indicates the additional predicted answer of the additional machine learning model output is a machine learning model generated prediction. 
 
   
     
     
         14 . The method of  claim 12 , further comprising:
 in response to receiving the search query:
 causing search results, that correspond to search resources that are responsive to the search query, to be rendered at the client device, 
 wherein transmitting, to the client device, the machine learning model output, occurs after causing the search results to be rendered. 
   
     
     
         15 . The method of  claim 14 , wherein when the machine learning model output is rendered in the interface, the search results are rendered, in the interface, positionally below rendering of the machine learning model output in the interface. 
     
     
         16 . The method of  claim 12 , wherein the machine learning model output further includes an interactive field, and further comprising:
 receiving, via the interactive field and after rendering of the machine learning model output, additional user interface input that varies from the search query;   in response to receiving the additional user interface input:
 processing the additional user interface input, using the machine learning model, to generate an additional predicted answer to the additional user interface input; and 
 causing the machine learning model output to be updated to include the generated additional predicted answer. 
   
     
     
         17 . The method of  claim 16 , further comprising:
 in response to receiving the search query:
 causing search results, that correspond to search resources that are responsive to the search query, to be rendered at the client device,. 
   
     
     
         18 . The method of  claim 12 , further comprising:
 in response to receiving the search query and in response to selecting the machine learning model:
 causing a prompt to be rendered, at the client device, that informs the user that the machine learning model output will be subsequently provided,
 wherein transmitting, to the client device, the machine learning model output, occurs after causing the prompt to be rendered. 
 
   
     
     
         19 . A system, comprising:
 a network interface;   memory storing instructions;   one or more processors operable to execute the instructions to:
 receive, via the network interface, a search query submitted by a client device of a user in response to user interface input that is provided by the user at the client device; 
 determine that search resources that are responsive to the search query fail to satisfy one or more criteria; 
 in response to receiving the search query and in response to determining that search resources that are responsive to the search query fail to satisfy the one or more criteria:
 process the search query using a machine learning model to generate a predicted answer to the search query; and 
 transmit, to the client device and in response to receiving the search query, machine learning model output that includes the predicted answer that is generated using the machine learning model, 
 wherein transmitting the machine learning model output causes the machine learning model output to be rendered, in an interface at the client device, with an indication that indicates the predicted answer of the machine learning model output is a machine learning model generated prediction.