US-12626166-B2 - Predicting the need for XAI in artificial intelligence systems

US12626166B2US 12626166 B2US12626166 B2US 12626166B2US-12626166-B2

Abstract

A method, computer program, and computer system are provided for selecting an artificial intelligence algorithm. Data corresponding to a user interaction with an artificial intelligence system is received. A need by the user for an explanation associated with the artificial intelligence system is identified based on the received data corresponding to the user interaction with the artificial intelligence system. An artificial intelligence algorithm is selected from among a plurality of artificial intelligence algorithms. The selected artificial intelligence algorithm includes an explainable artificial intelligence component. A model is created for the the selected artificial intelligence algorithm corresponding to the received data.

Inventors

VAGNER FIGUEREDO DE SANTANA
Ana Fucs
Vinicius Costa Villas Boas Segura
Daniel Brugnaro De Moraes
Renato Fontoura de Gusmao Cerqueira

Assignees

INTERNATIONAL BUSINESS MACHINES CORPORATION

Dates

Publication Date: 20260512
Application Date: 20220615

Claims (17)

1 . A computer-implemented method of selecting an artificial intelligence algorithm, executable by a processor, comprising: receiving received data corresponding to a user interaction in a user interface with an artificial intelligence system, wherein the received data includes a dataset selected by a user in the user interface, wherein the user interaction includes an event and a target; creating embeddings based on the received data and a usage graph used to train a model training module, wherein the usage graph comprises events and targets of users interacting with the artificial intelligence system; calculating a similarity value between the embeddings; identifying a need by the user for an explanation associated with the artificial intelligence system based on the similarity value; selecting an artificial intelligence algorithm from among a plurality of artificial intelligence algorithms, wherein the selected artificial intelligence algorithm includes an explainable artificial intelligence component; creating a model for the selected artificial intelligence algorithm corresponding to the received data; and providing feedback related to the created model back to the user through the user interface.
2 . The computer-implemented method of claim 1 , wherein the artificial intelligence algorithm is selected based on previous user interactions with the artificial intelligence system selecting an artificial intelligence algorithm having an explainable artificial intelligence component.
3 . The computer-implemented method of claim 1 , further comprising training the model based on measuring a similarity value between the received data and data corresponding to previous user interactions having a need for an explanation associated with the artificial intelligence system.
4 . The computer-implemented method of claim 1 , wherein the similarity value corresponds to a cosine similarity value between vectors associated with the created embeddings.
5 . The computer-implemented method of claim 1 , further comprising providing an explanation of the artificial intelligence algorithm to the user.
6 . The computer-implemented method of claim 1 , wherein the data corresponding to the user interaction with the artificial intelligence system includes data corresponding to an identifier, a timestamp, an event, a target, and metadata.
7 . A computer system for selecting an artificial intelligence algorithm, the computer system comprising: one or more computer-readable storage media configured to store computer program code; and one or more computer processors configured to access said computer program code and operate as instructed by said computer program code, said computer program code including: receiving code configured to cause the one or more computer processors to receive received data corresponding to a user interaction in a user interface with an artificial intelligence system, wherein the received data includes a dataset selected by a user in the user interface, wherein the user interaction includes an event and a target; creating code configured to cause the one or more computer processors to create embeddings based on the received data and a usage graph used to train a model training module, wherein the usage graph comprises events and targets of users interacting with the artificial intelligence system; calculating code configured to cause the one or more computer processors to calculate a similarity value between the embeddings; identifying code configured to cause the one or more computer processors to identify a need by the user for an explanation associated with the artificial intelligence system based on the similarity value; selecting code configured to cause the one or more computer processors to select an artificial intelligence algorithm from among a plurality of artificial intelligence algorithms, wherein the selected artificial intelligence algorithm includes an explainable artificial intelligence component; creating code configured to cause the one or more computer processors to create a model for the selected artificial intelligence algorithm corresponding to the received data; and providing feedback related to the created model back to the user through the user interface.
8 . The computer system of claim 7 , wherein the artificial intelligence algorithm is selected based on previous user interactions with the artificial intelligence system selecting an artificial intelligence algorithm having an explainable artificial intelligence component.
9 . The computer system of claim 7 , further comprising training code configured to cause the one or more computer processors to train the model based on measuring a similarity value between the received data and data corresponding to previous user interactions having a need for an explanation associated with the artificial intelligence system.
10 . The computer system of claim 7 , wherein the similarity value corresponds to a cosine similarity value between vectors associated with the created embeddings.
11 . The computer system of claim 7 , further comprising providing code configured to cause the one or more computer processors to provide an explanation of the artificial intelligence algorithm to the user.
12 . The computer system of claim 7 , wherein the data corresponding to the user interaction with the artificial intelligence system includes data corresponding to an identifier, a timestamp, an event, a target, and metadata.
13 . A computer readable medium having stored thereon a computer program for selecting an artificial intelligence algorithm, the computer program configured to cause one or more computer processors to: receive received data corresponding to a user interaction in a user interface with an artificial intelligence system, wherein the received data includes a dataset selected by a user in the user interface, wherein the user interaction includes an event and a target; create embeddings based on the received data and a usage graph used to train a model training module, wherein the usage graph comprises events and targets of users interacting with the artificial intelligence system; calculate a similarity value between the embeddings; identify a need by the user for an explanation associated with the artificial intelligence system based on the similarity value; select an artificial intelligence algorithm from among a plurality of artificial intelligence algorithms, wherein the selected artificial intelligence algorithm includes an explainable artificial intelligence component; create a model for the selected artificial intelligence algorithm corresponding to the received data; and providing feedback related to the created model back to the user through the user interface.
14 . The computer readable medium of claim 13 , wherein the artificial intelligence algorithm is selected based on previous user interactions with the artificial intelligence system selecting an artificial intelligence algorithm having an explainable artificial intelligence component.
15 . The computer readable medium of claim 13 , wherein the computer program is further configured to cause the one or more computer processors to train the model based on measuring a similarity value between the received data and data corresponding to previous user interactions having a need for an explanation associated with the artificial intelligence system.
16 . The computer readable medium of claim 13 , wherein the similarity value corresponds to a cosine similarity value between vectors associated with the created embeddings.
17 . The computer readable medium of claim 13 , wherein the computer program is further configured to cause the one or more computer processors to provide an explanation of the artificial intelligence algorithm to the user.

Description

FIELD This disclosure relates generally to field of machine learning, and more particularly to explainable artificial intelligence (XAI). BACKGROUND Explainable artificial intelligence (XAI) is a technique to generate an explanation for an artificial intelligence (AI) model's output or behavior. Each AI model has explanation evaluation criteria, which are desirable properties that an AI explanation should be judged by. These include the model being complete, generalizable, compact, stable, etc. SUMMARY Embodiments relate to a method, system, and computer readable medium for selecting an artificial intelligence algorithm. According to one aspect, a method for selecting an artificial intelligence algorithm is provided. The method may include receiving data corresponding to a user interaction with an artificial intelligence system. A need by the user for an explanation associated with the artificial intelligence system is identified based on the received data corresponding to the user interaction with the artificial intelligence system. An artificial intelligence algorithm is selected from among a plurality of artificial intelligence algorithms. The selected artificial intelligence algorithm includes an explainable artificial intelligence component. A model is created for the selected artificial intelligence algorithm corresponding to the received data. According to another aspect, a computer system for selecting an artificial intelligence algorithm is provided. The computer system may include one or more processors, one or more computer-readable memories, one or more computer-readable tangible storage devices, and program instructions stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, whereby the computer system is capable of performing a method. The method may include receiving data corresponding to a user interaction with an artificial intelligence system. A need by the user for an explanation associated with the artificial intelligence system is identified based on the received data corresponding to the user interaction with the artificial intelligence system. An artificial intelligence algorithm is selected from among a plurality of artificial intelligence algorithms. The selected artificial intelligence algorithm includes an explainable artificial intelligence component. A model is created for the selected artificial intelligence algorithm corresponding to the received data. According to yet another aspect, a computer readable medium for selecting an artificial intelligence algorithm is provided. The computer readable medium may include one or more computer-readable storage devices and program instructions stored on at least one of the one or more tangible storage devices, the program instructions executable by a processor. The program instructions are executable by a processor for performing a method that may accordingly include receiving data corresponding to a user interaction with an artificial intelligence system. A need by the user for an explanation associated with the artificial intelligence system is identified based on the received data corresponding to the user interaction with the artificial intelligence system. An artificial intelligence algorithm is selected from among a plurality of artificial intelligence algorithms. The selected artificial intelligence algorithm includes an explainable artificial intelligence component. A model is created for the selected artificial intelligence algorithm corresponding to the received data. According to one or more aspects, the artificial intelligence algorithm is selected based on previous user interactions with the artificial intelligence system selecting an artificial intelligence algorithm having an explainable artificial intelligence component. According to one or more aspects, the method may further include training the model based on measuring a similarity value between the received data and data corresponding to previous user interactions having a need for an explanation associated with the artificial intelligence system. According to one or more aspects, identifying the need by the user for the explanation associated with the artificial intelligence system includes creating embeddings based on the received data corresponding to the user interaction with the artificial intelligence system, calculating a similarity value between the embeddings, and determining the need for the explanation based on the calculated similarity value. According to one or more aspects, the similarity value corresponds to a cosine similarity value between vectors associated with the created embeddings. According to one or more aspects, the method may further include providing an explanation of the artificial intelligence algorithm to the user. According to one or more aspects, the data corresponding to the user interaction with the artificial intelligence syst