# US 12,602,413 B1 · Managing generative artificial intelligence (AI) model outputs using explainability reports

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,602,413 B1
- Document ID: US12602413B1
- Status: Granted
- Kind: B1
- Application number: 19/191,776
- Filed: 2025-04-28
- Granted: 2026-04-14
- Earliest claimed: 2025-04-28
- Assignees: Dell Products L.P.
- Applicants: Dell Products L.P.
- Inventors: Ofir Ezrielev; Hanna Yehuda; Anne-Marie Mcreynolds
- Web page: https://patentgrep.com/patent/US12602413B1

## CPC

- G06F 16/33295: {in dialogue systems}
- G06F 16/3328: {using graphical result space presentation or visualisation}
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/047: Probabilistic or stochastic networks
- G06N 3/0475: Generative networks
- G06N 3/08: Learning methods
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 5/041: {Abduction}
- G06N 5/045: Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Methods and systems for providing computer-implemented services using generative AI models are disclosed. To do so, a prompt may be obtained for a generative AI model of the generative AI models. The prompt and the generative AI model may be used to obtain a first output. An output explainability process may be performed using at least the prompt, the first output, and the generative AI model to obtain an explainability report for the first output. The explainability report may indicate relationships between first information elements of the prompt and second information elements of the first output. The relationships may be based, at least in part, on a second output generated by the generative AI model using a modified prompt based on the prompt. The first output and the explainability report may be provided to a downstream consumer as part of providing computer-implemented services to the downstream consumer.

## Claims

### Claim 1

A method for providing computer-implemented services using generative artificial intelligence (AI) models, the method comprising:
obtaining a prompt for a generative AI model of the generative AI models;
obtaining, using the prompt and the generative AI model, a first output from the generative AI model;
performing, using at least the prompt, the first output, and the generative AI model, an output explainability process to obtain an explainability report for the first output, the explainability report indicating relationships between first information elements of the prompt and second information elements of the first output, and the relationships are based, at least in part, on a second output generated by the generative AI model using a modified prompt and the modified prompt being based on the prompt; and
providing the first output and the explainability report to a downstream consumer as part of providing computer-implemented services to the downstream consumer.

### Claim 2 (depends on 1)

The method of claim 1, wherein performing the output explainability process comprises:
performing, using the prompt, a prompt modification process to obtain the modified prompt;
obtaining, using the modified prompt and the generative AI model, the second output from the generative AI model;
obtaining, using the second output, third information elements of the second output;
comparing the second information elements of the first output and the third information elements of the second output to obtain a difference; and
populating, based on the difference, the explainability report with at least a first relationship of the relationships.

### Claim 3 (depends on 2)

The method of claim 2, wherein performing the prompt modification process comprises:
obtaining, using the prompt and the first output, the first information elements and the second information elements; and
performing at least one action selected from a list of actions consisting of:
removing a first information element of the first information elements to obtain the modified prompt; and
replacing the first information element with a replacement information element to obtain the modified prompt.

### Claim 4 (depends on 3)

The method of claim 3, wherein performing the prompt modification process further comprises:
in an instance of the performing where the at least one action comprises replacing the first information element:
prior to replacing the first information element:
obtaining, based on the first information element and an information element replacement schema, the replacement information element.

### Claim 5 (depends on 3)

The method of claim 3, wherein performing the prompt modification process further comprises:
in an instance of the performing where the at least one action comprises replacing the first information element:
prior to replacing the first information element:
displaying, via a graphical user interface (GUI), at least the first information elements to a user; and
obtaining, via an interaction with the GUI, feedback from the user, the feedback comprising the replacement information element.

### Claim 6 (depends on 5)

The method of claim 5, wherein displaying the at least the first information elements to the user comprises:
obtaining a first graphical representation of the first information elements, the first graphical representation comprising:
a first set of nodes that correspond to the first information elements, and
first edges that indicate relationships between the first set of the nodes; and
populating the GUI with the first graphical representation.

### Claim 7 (depends on 5)

The method of claim 5, wherein comparing the second information elements of the first output and the third information elements of the second output comprises:
obtaining a second graphical representation of the second information elements, the second graphical representation comprising:
a second set of nodes that correspond to the second information elements, and
second edges that represent relationships between the second set of the nodes;
obtaining a third graphical representation of the third information elements, the third graphical representation comprising:
a third set of nodes that correspond to the third information elements, and
third edges that represent relationships between the third set of the nodes; and
identifying, using the second graphical representation and the third graphical representation, one or more nodes of the second set of the nodes that is not present in the third graphical representation.

### Claim 8 (depends on 7)

The method of claim 7, wherein the first output comprises a second information element and the second output does not comprise the second information element.

### Claim 9 (depends on 8)

The method of claim 8, wherein the second output comprises a third information element and the first output does not comprise the third information element.

### Claim 10 (depends on 7)

The method of claim 7, wherein the second graphical representation and the third graphical representation are displayed to the user via the GUI.

### Claim 11 (depends on 1)

The method of claim 1, wherein performing the output explainability process comprises:
iteratively perturbating information content of the prompt to obtain perturbated prompts;
evaluating the perturbated prompts using the generative AI model to obtain perturbated outputs; and
using the perturbated prompts and the perturbated outputs to identify the relationships between the first information elements and the second information elements.

### Claim 12 (depends on 1)

The method of claim 1, wherein the generative AI model is a large language model (LLM).

### Claim 13

A non-transitory machine-readable medium having instructions stored therein, which when executed by a processor, cause the processor to perform operations for providing computer-implemented services using generative artificial intelligence (AI) models, the operations comprising:
obtaining a prompt for a generative AI model of the generative AI models;
obtaining, using the prompt and the generative AI model, a first output from the generative AI model;
performing, using at least the prompt, the first output, and the generative AI model, an output explainability process to obtain an explainability report for the first output, the explainability report indicating relationships between first information elements of the prompt and second information elements of the first output, and the relationships are based, at least in part, on a second output generated by the generative AI model using a modified prompt and the modified prompt being based on the prompt; and
providing the first output and the explainability report to a downstream consumer as part of providing computer-implemented services to the downstream consumer.

### Claim 14 (depends on 13)

The non-transitory machine-readable medium of claim 13, wherein performing the output explainability process comprises:
performing, using the prompt, a prompt modification process to obtain the modified prompt;
obtaining, using the modified prompt and the generative AI model, the second output from the generative AI model;
obtaining, using the second output, third information elements of the second output;
comparing the second information elements of the first output and the third information elements of the second output to obtain a difference; and
populating, based on the difference, the explainability report with at least a first relationship of the relationships.

### Claim 15 (depends on 14)

The non-transitory machine-readable medium of claim 14, wherein performing the prompt modification process comprises:
obtaining, using the prompt and the first output, the first information elements and the second information elements; and
performing at least one action selected from a list of actions consisting of:
removing a first information element of the first information elements to obtain the modified prompt; and
replacing the first information element with a replacement information element to obtain the modified prompt.

### Claim 16 (depends on 15)

The non-transitory machine-readable medium of claim 15, wherein performing the prompt modification process further comprises:
in an instance of the performing where the at least one action comprises replacing the first information element:
prior to replacing the first information element:
obtaining, based on the first information element and an information element replacement schema, the replacement information element.

### Claim 17

A data processing system, comprising:
a processor; and
a memory coupled to the processor to store instructions, which when executed by the processor, cause the processor to perform operations for providing computer-implemented services using generative artificial intelligence (AI) models, the operations comprising:
obtaining a prompt for a generative AI model of the generative AI models;
obtaining, using the prompt and the generative AI model, a first output from the generative AI model;
performing, using at least the prompt, the first output, and the generative AI model, an output explainability process to obtain an explainability report for the first output, the explainability report indicating relationships between first information elements of the prompt and second information elements of the first output, and the relationships are based, at least in part, on a second output generated by the generative AI model using a modified prompt and the modified prompt being based on the prompt; and
providing the first output and the explainability report to a downstream consumer as part of providing computer-implemented services to the downstream consumer.

### Claim 18 (depends on 17)

The data processing system of claim 17, wherein performing the output explainability process comprises:
performing, using the prompt, a prompt modification process to obtain the modified prompt;
obtaining, using the modified prompt and the generative AI model, the second output from the generative AI model;
obtaining, using the second output, third information elements of the second output;
comparing the second information elements of the first output and the third information elements of the second output to obtain a difference; and
populating, based on the difference, the explainability report with at least a first relationship of the relationships.

### Claim 19 (depends on 18)

The data processing system of claim 18, wherein performing the prompt modification process comprises:
obtaining, using the prompt and the first output, the first information elements and the second information elements; and
performing at least one action selected from a list of actions consisting of:
removing a first information element of the first information elements to obtain the modified prompt; and
replacing the first information element with a replacement information element to obtain the modified prompt.

### Claim 20 (depends on 19)

The data processing system of claim 19, wherein performing the prompt modification process further comprises:
in an instance of the performing where the at least one action comprises replacing the first information element:
prior to replacing the first information element:
obtaining, based on the first information element and an information element replacement schema, the replacement information element.