Information processing system

Abstract

To indicate information on risks associated with AI.SOLUTION: An information processing system comprises a distributed network for realizing a distributed ledger. The distributed ledger stores one transaction data related to attribute information indicating one classification related to one learned model among a plurality of classifications of risks associated with AI.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to the technical field of information processing systems.

Background Art

[0002] As this type of system, for example, a data management system has been proposed that includes a relearning data ledger synchronized using distributed ledger technology and can retrospectively verify whether a machine learning model is appropriate (see Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The use of AI (Artificial Intelligence) including machine learning models is increasing. On the other hand, the risks brought about by the AI used are often unclear.

[0005] The present invention has been made in view of the above circumstances, and an object thereof is to propose an information processing system that can show information regarding risks brought about by AI.

Means for Solving the Problems

[0006] An information processing system according to an aspect of the present invention includes a distributed network for realizing a distributed ledger, and in the distributed ledger, one transaction data regarding attribute information indicating one classification related to one learned model among a plurality of classifications regarding risks brought about by AI is stored.

Brief Description of the Drawings

[0007]

Figure 1

Figure 2

Embodiments for Carrying Out the Invention

[0008] Embodiments related to the information processing system will be described with reference to FIGS. 1 and 2. FIG. 1 is a block diagram showing the configuration of the information processing system 1. FIG. 2 is a conceptual diagram showing the concept of the distributed ledger 4122.

[0009] In FIG. 1, the information processing system 1 includes a management server 10, a database 20, a terminal 30, and a distributed network 40. The distributed network 40 is a network for realizing a distributed ledger. In this embodiment, a blockchain is cited as an example of the distributed ledger. The terminal 30 may be connected to the management server 10 via a wide area network such as the Internet.

[0010] The distributed network 40 has nodes 41, 42, 43, and 44. Note that the number of nodes included in the distributed network 40 is not limited to "4". The node 41 may have a processor 411 and a memory 412. A computer program 4121 may be stored in the memory 412. The processor 411 may execute the processing to be performed by the node 41 together with the memory 412 in which the computer program 4121 is stored (in other words, together with the memory 412 and the computer program 4121 stored in the memory 412). For example, by executing the computer program 4121, a logical functional block for executing the processing to be performed by the node 41 may be realized in the processor 411.

[0011] In the memory 412 of the node 41, a distributed ledger 4122 realized by the distributed network 40 (in other words, constructed within the distributed network 40) may be stored. As shown in FIG. 2, one or more transactions Tx may be stored in the distributed ledger 4122. Note that the configurations of the nodes 42, 43, and 44 may be the same as that of the node 41.

[0012] The management server 10 may include a processor 11 and a memory 12. A computer program 121 may be stored in the memory 12. The processor 11 may execute the processing that the management server 10 should perform, together with the memory 12 in which the computer program 121 is stored (in other words, together with the memory 12 and the computer program 121 stored in the memory 12). For example, by executing the computer program 121, a logical functional block for executing the processing that the management server 10 should perform may be realized in the processor 11.

[0013] The management server 10 may provide the terminal 30 with an application for viewing the distributed ledger realized by the distributed network 40 and an application for accessing the database 20. For this reason, the management server 10 may be referred to as an application server.

[0014] The database 20 may store a trained model generated by machine learning. For example, the trained model may be a trained model applicable to a navigation device. Note that the trained model is not limited to a trained model applicable to at least one of the automatic driving system and the navigation device. Note that a transaction regarding information indicating the trained model registered in the database 20 may be stored in the distributed ledger 4122, or a distributed ledger different from the distributed ledger 4122. Note that a distributed ledger different from the distributed ledger 4122 may be realized by the distributed network 40, or may be realized by a distributed network different from the distributed network 40. Note that the trained model may not be registered in the database 20. In this case, the trained model may be registered in the distributed ledger.

[0015] The operation of the information processing system 1 when one trained model is registered will be described. Hereinafter, the case where the registration of one trained model is performed via the terminal 30 will be described. The operator of the terminal 30 may instruct the management server 10 to register one trained model via the terminal 30 (for example, using an application provided by the management server 10). At this time, the operator may also instruct the management server 10 to register the attribute information indicating one classification related to one trained model among a plurality of classifications regarding the risks brought about by AI.

[0016] The plurality of classifications regarding the risks brought about by AI may include, for example, at least one of "unacceptable risk", "high risk", "limited risk", and "minimum risk". "Unacceptable risk" may mean, for example, a risk that directly threatens people. "High risk" may mean, for example, a risk that affects people. "Limited risk" may mean that there is no serious risk, but there is an obligation to clarify that AI is acting. "Minimum risk" may mean that there is almost no risk or no risk.

[0017] For example, the processor 11 of the management server 10 may register one learned model in the database 20 and generate a transaction regarding information indicating the one learned model (for example, an identification number related to the one learned model, a hash value related to the one learned model, etc.). The processor 11 may further generate a transaction regarding attribute information indicating one classification related to the one learned model. Note that the information indicating the one learned model and the attribute information may be included in one transaction.

[0018] For example, the processor 411 of the node 41 included in the distributed network 40 may store a transaction regarding attribute information in the distributed ledger 4122 implemented by the distributed network 40. A transaction regarding information indicating one learned model may be stored in the distributed ledger by one node included in the distributed network 40 (for example, the node 41) or by one node included in a distributed network different from the distributed network 40.

[0019] Next, the operation of the information processing system 1 when one registered learned model is used will be described. Hereinafter, the case of using one learned model via the terminal 30 will be described. An operator of the terminal 30 may transmit utilization request information for requesting the use of one learned model to the management server 10 via the terminal 30 (for example, using an application provided by the management server 10).

[0020] The processor 11 of the management server 10 that has received the utilization request information may read out attribute information indicating one classification related to the one learned model from the distributed ledger 4211. Thereafter, the processor 11 may transmit display information for displaying information for using the one learned model to the terminal 30. The terminal 30 that has received the display information may display information for using the one learned model on, for example, a screen related to the application.

[0021] Here, the information for using a single pre-trained model includes information regarding a single classification related to the single pre-trained model. The information regarding a single classification may include, for example, a name indicating the classification (e.g., "unacceptable risk", "high risk", "limited risk", or "minimal risk"), and at least one of a number and a symbol indicating the classification. The information regarding a single classification may include information indicating the usage mode of the single pre-trained model. The information indicating the usage mode of the single pre-trained model may include, for example, at least one of "prohibited from use", "usable under predetermined conditions", and "usable without restrictions". The information regarding a single classification may include the terms of use related to the single pre-trained model. Note that the content of the terms of use may vary according to a single classification (i.e., the classification related to the risk brought about by AI). Note that when the information regarding a single classification includes the terms of use related to the single pre-trained model, consent to the terms of use may be set as a condition for using the single pre-trained model. Note that a plurality of terms of use corresponding to a plurality of classifications may be stored in the terminal 30. In this case, the terminal 30 may select, based on a single classification related to the single pre-trained model, one term of use corresponding to the single classification from among the plurality of terms of use. Then, the terminal 30 may display, for example, the selected terms of use on a screen related to the application.

[0022] (Technical effect) In the information processing system 1, a transaction regarding attribute information indicating a single classification related to a single pre-trained model is stored in the distributed ledger 4122. That is, in the distributed ledger 4122, classifications related to the risks brought about by AI for each pre-trained model are registered. Therefore, according to the information processing system 1, it is possible to indicate information regarding the risks brought about by AI.

[0023] In the information processing system 1, when there is a request to use a single learned model, information for using a single learned model including information regarding a single classification related to the single learned model may be transmitted, for example, from the management server 10 to the terminal 30. With such a configuration, before the single learned model is actually used, it is possible to show, for example, information regarding risks brought about by AI to the operator of the terminal 30.

[0024] In the information processing system 1, a distributed ledger 4122 that is extremely difficult to tamper with stores classifications regarding risks brought about by AI for each learned model. For this reason, the information processing system 1 can show highly reliable information regarding risks brought about by AI. In addition, when a plurality of classifications regarding risks brought about by AI are revised or the like, the classification related to a single learned model may be changed or added. In this case, the distributed ledger 4122 may store a transaction related to attribute information indicating the classification related to the single learned model before the change and a transaction related to attribute information indicating the classification related to the single learned model after the change. With such a configuration, it is possible to track the change history of the classification related to a single learned model.

[0025] Aspects of the invention derived from the embodiments described above will be described below.

[0026] An information processing system according to one aspect of the invention includes a distributed network for realizing a distributed ledger, and the distributed ledger stores one transaction data related to attribute information indicating one classification related to a single learned model among a plurality of classifications regarding risks brought about by AI.

[0027] When the information processing system receives a usage request for the one learned model, it may include output means for outputting information regarding the one classification by reading the one transaction data from the distributed ledger. Here, the information regarding the one classification may be the usage agreement related to the one learned model. In the above-described embodiment, "management server 10" corresponds to an example of the "output means".

[0028] The present invention is not limited to the above-described embodiments, and can be appropriately changed without departing from the gist or idea of the invention that can be read from the claims and the entire specification, and an information processing system accompanied by such a change is also included in the technical scope of the present invention.

Explanation of Reference Numerals

[0029] 1... Information processing system, 10... Management server, 40... Distributed network

Claims

1. Comprising a distributed network for realizing a distributed ledger, wherein the distributed ledger stores one transaction data regarding attribute information indicating one classification related to one learned model among a plurality of classifications regarding risks brought about by AI. An information processing system characterized by the above.

2. When a utilization request for the one learned model is received, it comprises output means for outputting information regarding the one classification by reading out the one transaction data from the distributed ledger. The information processing system according to claim 1, characterized by the above.

3. The information regarding the one classification is the terms of use related to the one learned model. The information processing system according to claim 2, characterized by the above.

Images