Prompt chaining system for generative artificial intelligence systems
A method for processing text prompts includes identifying a set of elements in a text prompt that form a basis for a generative output. The method also includes identifying an element of the set of elements that satisfy a refinement condition. The method further includes updating the element based on the element satisfying the refinement condition. The method still further includes generating the generative output in accordance with updating the element.
1 . A method for processing text prompts, comprising:
receiving a prompt at a user interface for detecting an input that forms a basis for a generative output from a generative AI model that is trained to perform one or more tasks associated with refining the input provided for the generative AI model;
detecting the received prompt and processing the receive prompt by:
identifying a first element in the received prompt associated with a visual representation,
identifying a second element that is understated or lacks detail,
using semantics to identify an ambiguous word or phrase, and/or
identifying a pattern that deviates from an expected norm in a given scene description;
receiving, from the generative AI model, a prioritized list of keywords or phrases for refining the received prompt in response to processing the received prompt;
automatically prompting a user to select one or more keywords and/or one or more phrases from the prioritized list of keywords and phrases;
synthesizing a new prompt from the received prompt in response to the user selecting the one or more keywords and/or the one or more phrases; and
repeating prompt synthesis until the user is satisfied with the generative output from the generative AI model.
2 . The method of claim 1 , wherein the generative output is a visualization.
3 . The method of claim 1 , wherein the ambiguous word or phrase is based on historical data.
4 . The method of claim 1 , wherein the method further comprises receiving additional prompts until the user is satisfied with the generative output.
5 . An apparatus for processing text prompts, comprising:
a processor; and
a memory coupled with the processor and storing instructions operable, when executed by the processor, to cause the apparatus to:
receive a prompt at a user interface for detecting an input that forms a basis for a generative output from a generative AI model that is trained to perform one or more tasks associated with refining the input provided for the generative AI model;
detect the received prompt and processing the receive prompt by:
identifying a first element in the received prompt associated with a visual representation,
identifying a second element that is understated or lacks detail,
using semantics to identify an ambiguous word or phrase, and/or
identifying a pattern that deviates from an expected norm in a given scene description;
receive, from the generative AI model, a prioritized list of keywords or phrases for refining the received prompt in response to processing the received prompt;
automatically prompt a user to select one or more keywords and/or one or more phrases from the prioritized list of keywords and phrases;
synthesize a new prompt from the received prompt in response to the user selecting the one or more keywords and/or the one or more phrases; and
repeat prompt synthesis until the user is satisfied with the generative output from the generative AI model.
6 . The apparatus of claim 5 , wherein the generative output is a visualization.
7 . The apparatus of claim 5 , wherein the ambiguous word or phrase is based on historical data.
8 . The apparatus of claim 5 , wherein execution of the instructions further cause the apparatus to receive additional prompts until the user is satisfied with the generative output.
9 . A non-transitory computer-readable medium having program code recorded thereon for processing text prompts, the program code executed by a processor and comprising:
program code to receive a prompt at a user interface for detecting an input that forms a basis for a generative output from a generative AI model that is trained to perform one or more tasks associated with refining the input provided for the generative AI model;
detect the received prompt and processing the receive prompt by:
identifying a first element in the received prompt associated with a visual representation,
identifying a second element that is understated or lacks detail,
using semantics to identify an ambiguous word or phrase, and/or
identifying a pattern that deviates from an expected norm in a given scene description;
program code to receive, from the generative AI model, a prioritized list of keywords or phrases for refining the received prompt in response to processing the received prompt;
program code to automatically prompt a user to select one or more keywords and/or one or more phrases from the prioritized list of keywords and phrases;
program code to synthesize a new prompt from the received prompt in response to the user selecting the one or more keywords and/or the one or more phrases; and
program code to repeat prompt synthesis until the user is satisfied with the generative output from the generative AI model.
10 . The non-transitory computer-readable medium of claim 9 , wherein the generative output is a visualization.
11 . The non-transitory computer-readable medium of claim 9 , wherein the ambiguous word or phrase is based on historical data.
12 . The non-transitory computer-readable medium of claim 9 , wherein the program code further comprises program code to receive additional prompts until the user is satisfied with the generative output.