JP7655691B1 - Information processing device, information processing method, and program - Google Patents

Abstract

[Problem] To provide an information processing device etc. that suppresses a decrease in customer satisfaction in responding to customer inquiries. [Solution] In an information processing device 100, a reception unit 110 receives a method selection signal related to a response method from a caller terminal. A determination unit 120 determines whether the caller is requesting a response by a generative artificial intelligence or a response by a human based on the method selection signal. If a response generation unit 130 determines that the caller is requesting a response by a generative artificial intelligence, it receives the inquiry from the caller, selects one generative artificial intelligence from a plurality of generative artificial intelligences prepared in advance, causes the selected generative artificial intelligence to generate a response, and transmits the response generated by the generative artificial intelligence to the caller terminal. If a connection processing unit 140 determines that the caller is requesting a response by a human, it connects the caller terminal and the staff terminal so that they can communicate with each other. [Selected Figure] FIG.

Description

This disclosure relates to an information processing device, an information processing method, and a program.

The use of automated response systems is increasing in customer service operations at call centers and other facilities. There are also proposals to use AI (Artificial Intelligence) in automated response systems.

For example, according to Patent Document 1, a customer support system acquires voice information, extracts keywords using voice recognition processing, identifies the type of inquiry and response information, and responds via voice call using AI.

The method described in Patent Document 2 determines a response sentence based on the content of the customer's speech, determines the content of the business operation requested, and issues instructions to the business operation execution device.

JP 2022-032112 A JP 2019-144400 A

However, the above-mentioned technologies may not provide the response that customers intended. As a result, there is a risk that customer satisfaction will decrease.

In view of the above problems, the present disclosure aims to provide an information processing device and the like that suppresses a decline in customer satisfaction when responding to customer inquiries.

The information processing device according to the present disclosure has a reception unit, a determination unit, an answer generation unit, and a connection processing unit. The reception unit receives a method selection signal relating to the answer method from the caller terminal. The determination unit determines whether the caller is requesting an answer by a generative artificial intelligence or an answer by a human based on the method selection signal. If the answer generation unit determines that the caller is requesting an answer by a generative artificial intelligence, it receives an inquiry from the caller and selects one generative artificial intelligence from a plurality of generative artificial intelligences prepared in advance. The answer generation unit causes the selected generative artificial intelligence to generate an answer, and transmits the answer generated by the generative artificial intelligence to the caller terminal. If the connection processing unit determines that the caller is requesting an answer by a human, it connects the caller terminal and the staff terminal so that they can communicate with each other.

In the information processing method according to the present disclosure, a computer executes the following processes. The computer receives a method selection signal relating to the answering method from the caller terminal 1. The computer determines whether the caller is requesting an answer by a generative artificial intelligence or an answer by a human based on the method selection signal. If the computer determines that the caller is requesting an answer by a generative artificial intelligence, it receives an inquiry from the caller and selects one generative artificial intelligence from a plurality of generative artificial intelligences prepared in advance. The computer causes the selected generative artificial intelligence to generate an answer, and transmits the answer generated by the generative artificial intelligence to the caller terminal 1. If the computer determines that the caller is requesting an answer by a human, it connects the caller terminal 1 and the staff terminal 300 so that they can communicate with each other.

The program according to the present disclosure causes a computer to execute the following information processing method. The computer receives a method selection signal relating to the answering method from the caller terminal 1. The computer determines whether the caller is requesting an answer by a generative artificial intelligence or an answer by a human based on the method selection signal. If the computer determines that the caller is requesting an answer by a generative artificial intelligence, it receives an inquiry from the caller and selects one generative artificial intelligence from a plurality of generative artificial intelligences prepared in advance. The computer causes the selected generative artificial intelligence to generate an answer and transmits the answer generated by the generative artificial intelligence to the caller terminal 1. If the computer determines that the caller is requesting an answer by a human, it connects the caller terminal 1 and the staff terminal 300 so that they can communicate with each other.

The present disclosure provides an information processing device, an information processing method, and a program that suppress a decrease in customer satisfaction when responding to customer inquiries.

1 is a block diagram of an information processing system according to a first embodiment. 1 is a block diagram of an information processing apparatus according to a first embodiment. FIG. 11 is a diagram showing the contents of selection information. FIG. 2 is a block diagram illustrating an example of a hardware configuration of a computer. 3 is a flowchart of an information processing method according to the first embodiment. FIG. 2 is a first sequence diagram of the information processing method according to the first embodiment. FIG. 2 is a second sequence diagram of the information processing method according to the first embodiment. FIG. 11 is a block diagram of an information processing system according to a second embodiment. FIG. 11 is a block diagram of an information processing apparatus according to a second embodiment. FIG. 11 is a sequence diagram of an information processing method according to the second embodiment.

The present invention will be described below through embodiments of the invention, but the invention according to the claims is not limited to the following embodiments. Furthermore, not all of the configurations described in the embodiments are necessarily essential as means for solving the problems. For clarity of explanation, the following description and drawings have been omitted and simplified as appropriate. In addition, the same elements are given the same reference numerals in each drawing, and duplicate explanations have been omitted as necessary.

First Embodiment
(Information Processing System 10)
An information processing system 10 will be described with reference to Fig. 1. Fig. 1 is a block diagram of the information processing system 10 according to the first embodiment. The information processing system 10 receives an inquiry from a caller P1 and responds to the received inquiry. The information processing system 10 is, for example, a call center system that handles inquiries, orders, complaints, and the like from customers. Alternatively, the information processing system 10 may be, but is not limited to, a contact center system.

The information processing system 10 is communicatively connected to a caller terminal 1 used by a caller P1 via a first network N1. The caller terminal 1 is, for example, a mobile phone or a smartphone used by the caller P1. The caller terminal 1 may be a landline phone or a computer with communication capabilities. The first network N1 is, for example, a public communications network for landlines or mobile phones. The information processing system 10 mainly comprises an information processing device 100, a server 200, and a staff terminal 300.

The information processing device 100 is connected to the caller terminal 1 via the first network N1 so that they can communicate with each other. The information processing device 100 is, for example, a computer or server with a communication function. The information processing device 100 can receive a call from the caller terminal 1 and respond in a predetermined manner using an automated voice. That is, the information processing system 10 may have at least a part of the functionality of an IP-PBX (Internet protocol private branch exchange). The information processing device 100 may have at least a part of the functionality of an IVR (Interactive Voice Response). The information processing device 100 automatically responds to an incoming call from the caller terminal 1 and plays a preset response message. The information processing device 100 also causes the caller P1 to press a number button on the phone to send a tone signal. The information processing device 100 connects the staff terminal 300 and the caller terminal 1 so that they can talk to each other depending on the situation.

The information processing device 100 is communicatively connected to the server 200 and can access the language model held by the server 200. This allows the information processing device 100 to obtain, depending on the situation, a response to a prompt generated by the information processing device 100 from the server 200.

The information processing device 100 can be connected to the staff terminal 300 so that they can communicate with each other. Depending on the situation, the information processing device 100 relays between the caller terminal 1 and the staff terminal 300, allowing a call between the caller P1 and the staff P2. Note that relaying between the caller terminal 1 and the staff terminal 300 can also be said to transfer the line connected to the caller terminal 1 to the staff terminal 300.

The server 200 is communicatively connected to the information processing device 100. The server 200 has a first language model 210 and a second language model 220. The first language model 210 and the second language model 220 are generative AI (Artificial Intelligence) using a large-scale language model (LLM (Large Language Model)) that generates answers according to various user requests. Generative AI can also be called generative artificial intelligence. In this disclosure, the first language model 210 and the second language model 220 are different language models. Here, "different language models" includes cases where the types of language models themselves are different. Also, "different language models" includes cases where the types of language models are the same but the versions are different. The first language model 210 or the second language model 220 receives a prompt from the information processing device 100, generates an answer to the received prompt, and supplies the answer to the information processing device 100.

The information processing device 100 and the server 200 may be connected to each other via a wired or wireless connection so that they can communicate directly with each other. The information processing device 100 and the server 200 may be connected to each other via an Internet line. The server 200 may be a virtual entity provided in the cloud. In addition, in FIG. 1, one server has the first language model 210 and the second language model 220, but multiple servers may have the first language model 210 and the second language model 220. In addition, in FIG. 1, the server 200 has two generation AIs, the first language model 210 and the second language model 220, but it may have three or more language models.

The staff terminal 300 is a terminal used by staff member P2, and is communicatively connected to the information processing device 100. The staff terminal 300 is, for example, a computer with a communication function. The staff terminal 300 is set up so that it can communicate with caller P1 using caller terminal 1 via the information processing device 100.

(Information processing device 100)
Next, the information processing device 100 will be described with reference to Fig. 2. The information processing device 100 receives an outgoing call from the caller terminal 1 via the first network N1. The information processing device 100 responds to an inquiry from the caller P1 by receiving the outgoing call from the caller terminal 1. The information processing device 100 mainly includes a reception unit 110, a determination unit 120, a response generation unit 130, a connection processing unit 140, a communication unit 150, and a storage unit 160.

The reception unit 110 receives an outgoing call from the caller terminal 1. After the reception unit 110 establishes a connection with the caller terminal 1, it sends a predefined reception message to the caller terminal 1. The reception message includes a message that prompts the caller to specify a response method for the inquiry from the caller P1.

The reception unit 110 receives a method selection signal regarding the answering method from the caller terminal 1. The answering method is a method of answering an inquiry received from the caller P1. The answering method includes an answer by a generated AI and an answer by a human. The method selection signal regarding the answering method is a signal indicating which answering method the caller P1 desires from an answer by a generated AI and an answer by a human. This method selection signal is, for example, a tone signal that can be transmitted by the caller terminal 1. Note that the tone signal in this disclosure is a tone signal transmitted by a general telephone or a smartphone. The tone signal may also be referred to as DTMF (Dual-Tone Multi-Frequency).

The method selection signal may be a signal other than a tone signal. The method selection signal may be voice data that is an utterance by the caller P1. The method selection signal may be a signal that is text data input by the caller P1 and transmitted by the caller terminal 1. The method selection signal may also be a combination of these.

Based on the above-mentioned method selection signal, the determination unit 120 determines whether the caller P1 is requesting a response by the generated AI or a response by a human. Specifically, for example, when the method selection signal is set to a tone signal, the determination unit 120 detects the tone signal received from the caller terminal 1 and determines the type of the detected tone signal.

More specifically, for example, it is assumed that the tone signal requesting a human response corresponds to "1" on a push-button telephone line, and the tone signal requesting a response from the generation AI corresponds to "2" on a push-button telephone line. In this case, the determination unit 120 determines whether the method selection signal is "1", "2", or something other than these.

In the above-mentioned determination, the determination unit 120 refers to the selection information 161 stored in the storage unit 160. Here, the selection information 161 includes information linking the type of tone signal with the response method.

The determination unit 120 may be configured to be able to select which of the multiple generation AIs possessed by the information processing system 10 to request an answer from. In this case, for example, the answer from the first language model 210 may be set to "2", and the answer from the second language model 220 may be set to "3". In this case, the determination unit 120 determines whether the method selection signal is "1", "2", "3", or other than these. In this case, the selection information 161 stored in the memory unit 160 may also include information regarding the type of generation AI that will provide an answer in the information linking the type of tone signal and the answering method.

The determination of whether to request a response from the generated AI or a human response may be set by recognizing the speech of the caller P1. In this case, the determination unit 120 detects the speech of the caller P1 and determines from the content of the detected speech whether the caller P1 requests a response from the generated AI or a human response. Alternatively, the determination unit 120 may determine from the content of the speech of the caller P1 which of the multiple generated AIs possessed by the information processing system 10 to request a response from. In this case, the selection information 161 stored in the storage unit 160 may also include information regarding the type of generated AI that will respond in information linking the content of the speech of the caller P1 with the response method.

When the determination unit 120 determines that the caller P1 is requesting an answer from the generation AI, the answer generation unit 130 accepts an inquiry from the caller P1. In this case, the answer generation unit 130 transmits voice data to the caller terminal 1 to prompt the caller P1 to speak the content of the inquiry. Furthermore, when the answer generation unit 130 detects voice data that is the speech of the caller P1, it converts the accepted voice data into text data in order to recognize the content of the inquiry. Here, the answer generation unit 130 performs tokenization on the text data and generates a prompt from the content of the inquiry.

That is, the answer generation unit 130 converts the inquiry, which is the speech of the caller P1, into text, and generates a prompt for one generation AI to generate an answer based on the converted text. A prompt is an instruction sentence input to the generation AI. The prompt includes specific content for generating an answer to the inquiry of the caller P1.

The answer generation unit 130 also selects one generation AI from multiple generation AIs prepared in advance. That is, the answer generation unit 130 selects either the first language model 210 or the second language model 220 held by the server 200. The answer generation unit 130 makes this selection in response to either a method selection signal or the speech of the caller P1. For example, if the method selection signal includes the selection of the first language model 210 from the generation AIs, the answer generation unit 130 selects the first language model 210 in accordance with the method selection signal.

Alternatively, the answer generation unit 130 selects the generated AI from the contents of the data obtained by converting the speech of the sender P1 into text. In this case, the answer generation unit 130 refers to the selection information 161 stored in the storage unit 160. The selection information 161 in this case is information that links the text related to the content of the inquiry with the type of generated AI. The answer generation unit 130 compares the data obtained by converting the speech of the sender P1 into text with the text included in the selection information 161, and extracts the same text or text with a similar meaning. Furthermore, the answer generation unit 130 selects a language model that is closely connected to the speech of the sender P1 from the extracted results.

The answer generation unit 130 causes the selected generation AI to generate an answer, and transmits the answer generated by the generation AI to the caller terminal 1. That is, the answer generation unit 130 generates a prompt to generate an answer for the selected language model. The answer generation unit 130 supplies the generated prompt to either the first language model 210 or the second language model 220. The answer generation unit 130 receives the answer generated in response to the prompt.

The answer generation unit 130 also converts the answer generated by the generation AI into voice data and transmits it to the caller terminal 1. That is, the answer generation unit 130 converts the content of the answer received from the language model into voice data in order to convey it to the caller P1. The answer generation unit 130 then transmits the converted voice data to the caller terminal 1. At this time, the information processing device 100 transmits the voice data to the caller terminal 1 via the communication unit 150.

To realize the above-mentioned functions, the answer generation unit 130 has, as its main components, a signal conversion unit 131, a prompt generation unit 132, and a language model selection unit 133.

The signal conversion unit 131 converts text data into voice data. The signal conversion unit 131 also converts voice data into text data. For example, the signal conversion unit 131 converts voice data, which is an utterance by the caller P1, into text data. The signal conversion unit 131 also converts a response, which is text data received from the language model, into voice data.

The prompt generation unit 132 generates a prompt for one generation AI to generate an answer based on text including the inquiry content spoken by the caller P1.

The language model selection unit 133 selects one generation AI based on a method selection signal, which is a tone signal in the telephone function of the caller terminal 1 or voice data uttered by the caller. Alternatively, the language model selection unit 133 selects one generation AI based on the content of words included in the inquiry uttered by the caller P1. The language model selection unit 133 refers to the selection information 161 stored in the memory unit 160.

When the determination unit 120 determines that the caller P1 is requesting a human response, the connection processing unit 140 connects the caller terminal 1 and the staff terminal 300 so that they can communicate. In other words, when the method selection signal requests a human response, the connection processing unit 140 connects the line that receives the call from the caller terminal 1 to the staff terminal 300. The connection processing unit 140 relays between the caller terminal 1 and the staff terminal 300, so that the caller P1 and the staff P2 can communicate with each other.

The communication unit 150 is responsible for functions related to communication with the first network N1, communication with the server 200, and communication with the staff terminal 300. In other words, the communication unit 150 performs signal processing related to the transmission and reception of signals in these communication functions.

The storage unit 160 stores various information for implementing the functions of the information processing device 100. For example, the storage unit 160 stores a reception message that the reception unit 110 transmits to the caller terminal 1. The storage unit 160 also stores selection information 161. The selection information 161 is information that the answer generation unit 130 uses when selecting a generation AI.

The above describes the information processing device 100. With the above-mentioned configuration, the information processing device 100 can provide both human responses and responses from the generated AI in accordance with the customer's request. The information processing device 100 also selects a generated AI from multiple generated AIs according to the inquiry content, and provides the caller with an answer according to the inquiry content. Thus, the information processing device 100 can suppress a decrease in customer satisfaction in responding to customer inquiries.

(Selection information)
Next, the selection information 161 will be described with reference to Fig. 3. Fig. 3 is a diagram showing the contents of the selection information 161. The selection information 161 shown in Fig. 3 is information that links the type of language model with a plurality of words that are expected to be included in the query content.

Specifically, inquiry words linked to the "first language model" include "specs," "specifications," "lineup," "how to use," "what color," etc. In this case, the first language model 210 is a language model that has learned about inquiries and responses about products. Inquiry words linked to the "second language model" include "fault," "broken," "strange," "not working," "abnormal," etc. In this case, the second language model 220 is a language model that has learned about inquiries and responses about after-sales service.

The answer generation unit 130 compares the words included in the inquiry uttered by the sender P1 with the words included in the selection information 161. The answer generation unit 130 then selects the one that contains more words that are close to the content of the inquiry from the sender P1.

The selection information 161 is not limited to the above. In the information processing system 10, if the method selection signal includes content for selecting a generation AI, the selection information 161 may include, for example, information linking the type of tone signal with the type of generation AI. In addition, the selection information 161 may be able to be added, deleted, updated, etc., as appropriate.

(Hardware configuration example)
Fig. 4 is a block diagram illustrating an example of a hardware configuration of a computer. The above-mentioned information processing apparatus 100 may have the configuration shown in Fig. 4. The computer 1000 has a bus 1010, a processor 1020, a memory 1030, a storage device 1040, an input/output interface 1050, and a network interface 1060.

The bus 1010 is a data transmission path for the processor 1020, memory 1030, storage device 1040, input/output interface 1050, and network interface 1060 to transmit and receive data to and from each other. However, the method of connecting the processor 1020 and other components to each other is not limited to a bus connection.

The processor 1020 is a circuit that includes arithmetic devices such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit).

Memory 1030 is a main storage device realized using RAM (Random Access Memory) or the like.

Storage device 1040 is an auxiliary storage device such as a hard disk drive (HDD), a solid state drive (SSD), a flash memory, or a read only memory (ROM). Storage device 1040 stores programs for implementing the functions of the present disclosure.

The processor 1020 reads this program into the memory 1030 and executes it. This causes the processor 1020 to execute the function corresponding to this program. In other words, the program stored in the memory 1030 causes the computer 1000 to execute the functions of the present disclosure.

The input/output interface 1050 connects the computer 1000 to a specific input/output device. The input/output device is, for example, an input device such as a keyboard, an output device such as a display, or an input/output device in which a touch panel is superimposed on a display.

The network interface 1060 is an interface for connecting the computer 1000 to a specified communication network.

The computer 1000 has been described above, but in addition to the above configuration, the computer 1000 may have an information input device that allows the user to input various information to the computer 1000 through operation. The information input device is, for example, a keyboard, a mouse, or a touch panel. The computer 1000 may also have a display, a speaker, a vibration motor, or an LED (light-emitting diode) for showing various information to the user.

(Information Processing Method)
Next, a process executed by the information processing device 100 will be described with reference to Fig. 5. Fig. 5 is a flowchart of an information processing method according to the first embodiment. The process executed by the information processing device 100 in Fig. 5 starts, for example, when the information processing system 10 accepts an outgoing call from the caller terminal 1.

In step S11, the information processing device 100 transmits a reception message to the caller terminal 1 with which communication has been established. The reception message includes a message prompting the caller to transmit a method selection signal to select a response method. For example, the reception message may say, "Thank you for calling. If you would like to be answered by a staff member, press '1' and '#'. If you would like to be answered by the generating AI, press '2' and '#'."

In step S12, the reception unit 110 receives a method selection signal related to the answering method from the caller terminal 1. The reception unit 110 supplies information related to the received answering method to the determination unit 120.

In step S13, the determination unit 120 determines whether the caller P1 is requesting a response from the generated AI or a response from a human based on the method selection signal. More specifically, for example, the determination unit 120 determines whether the caller P1 is requesting a response from the generated AI. If the determination unit 120 determines that the caller P1 is requesting a response from the generated AI (step S13: YES), the information processing device 100 proceeds to step S14. On the other hand, if the determination unit 120 does not determine that the caller P1 is requesting a response from the generated AI (step S13: NO), the information processing device 100 proceeds to step S20.

In step S14, if the determination unit 120 determines that the caller P1 is requesting a response from the generation AI, the response generation unit 130 requests the content of the inquiry from the caller P1. That is, the response generation unit 130 transmits a message such as "Please tell us the content of your inquiry" as voice data to the caller terminal 1.

In step S15, the answer generation unit 130 accepts an inquiry from the sender P1. Upon accepting the inquiry, the signal conversion unit 131 of the answer generation unit 130 converts the accepted inquiry into text data. The language model selection unit 133 selects, from the text data, whether to have the first language model 210 or the second language model 220 answer the inquiry. In other words, the language model selection unit 133 selects one generation AI from multiple generation AIs prepared in advance.

In step S16, the prompt generation unit 132 determines whether or not the answer generation unit 130 should have the first language model 210 answer the inquiry based on the result of the selection by the language model selection unit 133. If the answer generation unit 130 determines that the first language model 210 should answer the inquiry (step S16: YES), the answer generation unit 130 proceeds to step S17. On the other hand, if the answer generation unit 130 does not determine that the first language model 210 should answer the inquiry (step S16: NO), the answer generation unit 130 proceeds to step S22.

In step S17, the answer generation unit 130 causes the selected generation AI to generate an answer. That is, the prompt generation unit 132 of the answer generation unit 130 sends a prompt to the first language model 210 to generate an answer to the inquiry.

In step S18, the answer generation unit 130 transmits the answer generated by the generation AI to the caller terminal 1. The answer generation unit 130 converts the answer generated by the first language model 210 or the second language model 220 into voice data, and transmits the converted voice data to the caller terminal 1. In this process, the information processing device 100 may transmit voice data to the caller terminal 1 saying, "If you are satisfied with the answer, please hang up the phone. If you have any further inquiries, please state your additional inquiry." By transmitting such voice data, the information processing device 100 can continue responding using the generation AI if the caller P1 wishes to continue the inquiry.

In step S19, the information processing device 100 judges whether or not to end the series of processes. The series of processes is ended when, for example, communication with the caller terminal 1 is ended. On the other hand, the series of processes is not ended when communication with the caller terminal 1 is not ended. When the information processing device 100 judges that the series of processes is ended (step S19: YES), the information processing device 100 ends the series of processes. On the other hand, when the information processing device 100 does not judge that the series of processes is ended (step S19: NO), the information processing device 100 returns to step S14 and requests the inquiry content again. Note that when the caller P1 repeats an inquiry, the information processing device 100 may record the history of the inquiry and the answer. In this case, the information processing device 100 may have a function of generating the contents of the prompt from the information of the recorded history. This allows the information processing device 100 to reduce the effort of the caller P1 in repeatedly making inquiries.

In step S13, if the determination unit 120 does not determine that the caller P1 is requesting a response from the generated AI (step S13: NO), it can be said that the caller P1 is requesting a response from a human. Therefore, in step S20, the connection processing unit 140 connects the caller terminal 1 and the staff terminal 300 so that they can communicate with each other.

In step S21, the connection processing unit 140 determines whether the call between the caller terminal 1 and the staff terminal 300 has ended. If it is not determined that the call between the caller terminal 1 and the staff terminal 300 has ended (step S21: NO), the connection processing unit 140 repeats step S21. On the other hand, if it is determined that the call between the caller terminal 1 and the staff terminal 300 has ended (step S21: YES), the information processing device 100 ends the series of processes.

In step S16, if the answer generation unit 130 does not determine that the first language model 210 should be made to answer (step S16: NO), the answer generation unit 130 proceeds to step S22. In step S22, the answer generation unit 130 makes the selected generation AI generate an answer. That is, the prompt generation unit 132 sends a prompt to the second language model 220 to generate an answer to the inquiry. After making the second language model 220 generate an answer, the answer generation unit 130 proceeds to step S18.

The above describes the processing executed by the information processing device 100. Through the above processing, the information processing device 100 finds out the contents of the inquiry from the caller P1, selects a generation AI from multiple generation AIs according to the contents of the inquiry, and generates an answer. Therefore, the information processing device 100 can suppress a decrease in customer satisfaction in responding to customer inquiries.

The process executed by the information processing device 100 is not limited to the above-mentioned configuration. For example, in step S11, the reception message sent by the reception unit 110 may be a message that accepts three options: "Answer by staff", "Answer by AI generated about the product", and "Answer by AI generated about after-sales service". In this case, the information processing device 100, for example, causes the first language model 210 to answer inquiries about products. The first language model 210 is a language model that has learned about inquiries and answers about products. The information processing device 100 causes the second language model 220 to answer inquiries about after-sales service. The second language model 220 is a language model that has learned about inquiries and answers about after-sales service.

In addition, in step S11, the reception unit 110 may include the working status of the staff in the reception message. The working status of the staff may include information indicating whether or not there is a staff member available to make a call. This allows the information processing device 100 to provide the caller P1 with information as to whether or not a call with the staff member P2 is immediately possible. Therefore, in this case, the information processing device 100 can encourage the caller P1 to make a smooth decision when selecting a response method.

In addition, in step S19, if the information processing device 100 does not determine that the series of processes is to be terminated (step S19: NO), the information processing device 100 may return to step S12 instead of step S14 and perform processing to accept the answering method again. This allows the information processing device 100 to change the setting to a human answer if the sender P1 is not satisfied with the answer of the generated AI.

Next, an example of the interaction between the caller terminal 1 and the information processing system 10 will be described with reference to FIG. 6. FIG. 6 is a first sequence diagram of the information processing method according to the first embodiment. The sequence diagram shown in FIG. 6 shows the interaction between the caller terminal 1 and the information processing system 10 when the caller P1 requests a response from the generation AI.

In step S101, the caller terminal 1 makes a call to the information processing system 10. That is, the caller terminal 1 transmits a call signal to the information processing system 10. The information processing device 100 of the information processing system 10 receives this signal and receives an incoming call.

In step S102, the information processing device 100 accepts the incoming call and transmits a reception message to the caller terminal 1. The reception message includes content that prompts the caller to select a response method.

In step S103, the sender P1 selects a reply method. This causes the sender terminal 1 to transmit a method selection signal to the information processing device 100. Note that in the sequence diagram shown in FIG. 6, the sender P1 selects a reply by the generation AI.

In step S104, the information processing device 100, which has received a method selection signal for selecting an answer by the generation AI, transmits a response message to the caller terminal 1 to request the contents of the inquiry.

In step S105, the caller P1 speaks the inquiry. The caller terminal 1 transmits voice data including the inquiry spoken by the caller P1 to the information processing device 100.

In step S106, the information processing device 100 selects a language model. As an example, it is assumed here that the information processing device 100 selects the first language model 210. The information processing device 100 generates a prompt to cause the first language model 210 to generate an answer.

In step S107, the information processing device 100 prompts the first language model 210 to generate an answer. The instruction to generate an answer includes the inquiry content uttered by the caller P1.

In step S108, the first language model 210 generates an answer to the prompt and provides it to the information processing device 100.

In step S109, the information processing device 100 generates voice data from the answer received from the first language model 210. Alternatively, the information processing device 100 may generate an answer sentence that includes the answer generated by the first language model 210, and then convert the answer sentence into voice data.

In step S110, the information processing device 100 transmits voice data including the response to the caller terminal 1.

Next, another example of the interaction between the caller terminal 1 and the information processing system 10 will be described with reference to FIG. 7. FIG. 7 is a second sequence diagram of the information processing method according to the first embodiment. The sequence diagram shown in FIG. 7 shows the interaction between the caller terminal 1 and the information processing system 10 when the caller P1 requests a human response. Note that the interaction in steps S101 and S102 is the same as in the sequence diagram shown in FIG. 6. Therefore, a description thereof will be omitted here.

In step S103, the sender P1 selects a reply method. This causes the sender terminal 1 to transmit a method selection signal to the information processing device 100. In the sequence diagram shown in FIG. 7, the sender P1 selects a human reply.

In step S201, the information processing device 100 receives a method selection signal for selecting a human response and connects the caller terminal 1 to the staff terminal 300.

In step S202, the information processing device 100 transmits a call signal to the staff terminal 300. Staff member P2, who uses the staff terminal 300, receives the call signal transmitted by the information processing device 100 and sets the staff terminal 300 to a state in which communication is possible. After this, the caller terminal 1 and the staff terminal 300 are in a state in which communication is possible.

The above describes the interaction between the caller terminal 1 and the information processing system 10. As described above, the information processing system 10 appropriately sets an answer based on a language model or an answer by a person in response to a method selection signal sent from the caller terminal 1. Furthermore, when an answer based on a generation AI is selected, the information processing system 10 selects one generation AI from multiple generation AIs to generate an answer.

The first embodiment has been described above. Note that the information processing system 10 is not limited to the above-mentioned configuration. The information processing device 100 can be connected to multiple caller terminals 1. Note that the number of lines that can be connected may be set in advance. The information processing system 10 can also include multiple staff terminals 300. In this case, the communication unit 150 recognizes the connection status of each of the multiple staff terminals 300, and connects to the caller terminal 1 according to the connection status. With the above-mentioned configuration, the information processing device 100 can respond using the generation AI even when the number of staff is small compared to the number of inquiries.

The generation AI of the server 200 may be multimodal compatible. In this case, the answer generation unit 130 may supply the voice data including the inquiry of the caller P1 to the server 200 without converting it into text data. In addition, when the generation AI outputs an answer using voice data, the answer generation unit 130 may transmit the voice data output by the generation AI to the caller terminal 1.

The information processing system 10 and the information processing device 100 have been described above. Each component of the information processing device 100 may be realized by dedicated hardware. In addition, a part or all of each component may be realized by a general-purpose or dedicated circuit, processor, etc., or a combination of these. These may be configured by a single chip, or may be configured by multiple chips connected via a bus. A part or all of each component may be realized by a combination of the above-mentioned circuits, etc., and a program. In addition, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an FPGA (field-programmable gate array), etc. may be used as a processor. In addition, at least a part of the functions of this embodiment may be provided in the form of IaaS (Infrastructure as a Service), PaaS (Platform as a Service), SaaS (Software as a Service), etc.

As described above, this embodiment can provide an information processing device 100, an information processing method, and a program that can suppress a decrease in customer satisfaction when responding to customer inquiries.

Second Embodiment
Next, a second embodiment will be described. Fig. 8 is a block diagram of an information processing system 10 according to the second embodiment. The information processing system 10 according to this embodiment differs from the information processing system 10 according to the first embodiment in that it has a reference information storage unit 400. The information processing system 10 according to this embodiment also differs from the information processing system 10 according to the first embodiment in that it is connected to the second network N2 in addition to the first network N1 so as to be able to make calls.

The information processing device 100 according to the second embodiment is connected to a second network N2 in addition to a first network N1. The first network N1 is a public communications network for landlines or mobile phones. On the other hand, the second network N2 is, for example, an Internet line. That is, the information processing device 100 can be connected to a caller terminal 1 so as to be able to make a call, and can also communicate with the caller terminal 1 by information communication means other than calls. More specifically, for example, the information processing device 100 can transmit information by any data, such as text data, image data, etc.

The information processing device 100 according to the second embodiment is communicatively connected to the reference information storage unit 400. The reference information storage unit 400 stores information related to an inquiry from the sender P1. For example, in a case where it is expected that the sender P1 will make an inquiry about a specific product, the reference information storage unit 400 stores information related to the product about which the inquiry is made. The reference information storage unit 400 can provide the stored information to the information processing device 100. When the information processing device 100 has the first language model 210 or the second language model 220 respond to the inquiry received from the sender P1, the information processing device 100 can generate a prompt by referring to the information in the reference information storage unit 400. That is, the information processing device 100 has a RAG function (Retrieval-Augmented Generation). Note that the information processing device 100 and the reference information storage unit 400 may be configured as an integrated unit. That is, the information processing device 100 may include the reference information storage unit 400.

Figure 9 is a block diagram of an information processing device according to the second embodiment. The information processing device 100 according to this embodiment differs from the information processing device 100 according to the first embodiment mainly in the configurations of the reception unit 110, the response generation unit 130, the connection processing unit 140, and the storage unit 160.

The reception unit 110 in this embodiment performs confirmation for the caller P1 to identify the customer. The purpose of identifying the customer is to send a response to the caller P1's inquiry to the caller terminal 1 as text data. The information processing system 10 may authenticate the caller terminal 1 or the caller P1 via the first network N1 and the second network N2. That is, the reception unit 110 may have a function of authenticating the caller P1.

The answer generation unit 130 according to this embodiment has a search function unit 134. The search function unit 134 is configured to realize the RAG function. The search function unit 134 refers to the contents of the inquiry of the sender P1, extracts information related to the contents of the inquiry from the reference information storage unit 400, and supplies the extracted information to the prompt generation unit 132. The prompt generation unit 132 generates a prompt including the reference information received from the search function unit 134 in addition to the contents of the inquiry of the sender P1. That is, the answer generation unit 130 searches a specified database for information related to the inquiry and generates a prompt using the search results. As a result, the first language model 210 or the second language model 220 of the server 200 generates an answer to the inquiry by referring to the reference information.

The answer generation unit 130 according to the present embodiment transmits text data to the sender terminal 1 via the second network N2. At this time, the answer generation unit 130 refers to the customer information 162 stored in the storage unit 160 and identifies the means for transmitting the text data to the sender P1 from the personal information of the sender P1. The means for transmitting the text data to the sender P1 is, for example, an email address. The email address of the sender P1 may be included in the voice data acquired by the information processing device 100 by the sender P1 speaking. The means for transmitting the text data to the sender P1 may be the account information of the sender P1 in a specified SNS. By the answer generation unit 130 transmitting the text data to the sender terminal 1, the information processing system 10 can convey a message to the sender P1 by voice data as well as by text data. The text data transmitted by the answer generation unit 130 may include information regarding the type of language model that generated the answer. Therefore, the information processing system 10 can respond with high customer satisfaction.

The memory unit 160 in this embodiment has customer information 162. The customer information 162 includes a unique identifier for the customer and information regarding a means of contacting the customer.

Next, with reference to FIG. 10, the exchange between the caller terminal 1 and the information processing system 10 will be described. FIG. 10 is a sequence diagram of the information processing method according to the second embodiment. The sequence diagram shown in FIG. 10 differs from the sequence diagram in FIG. 6 in the processing after step S108.

After receiving the answer from the first language model 210 in step S108, the information processing device 100 generates voice data and text data of the answer in step S301. The voice data is data to be transmitted to the caller terminal 1 via the first network N1. The text data is data to be transmitted to the caller terminal 1 via the second network N2. The content of the voice data and the content of the text data may be the same, or may differ in some parts. For example, the text data may include more detailed information than the voice data. The text data may include a specified URL (Uniform Resource Locator) related to the answer. The voice data may also include content explaining the difference from the text data. This allows the information processing device 100 to provide a more polite answer to the customer.

In step S302, the information processing system 10 transmits voice data regarding the response to the caller terminal 1 via the first network N1.

In step S303, the information processing system 10 transmits text data regarding the answer to the sender terminal 1 via the second network N2. That is, the answer generation unit 130 of the information processing device 100 transmits the answer generated by the generation AI to the sender terminal 1 as text data.

The second embodiment has been described above. In this embodiment, by using the RAG function in responding to customer inquiries, detailed responses can be provided quickly. This embodiment also allows responses to be provided using text data in addition to voice data. As a result, this embodiment can provide an information processing device 100, information processing method, and program that provide high customer satisfaction.

The above-mentioned program includes a set of instructions (or software code) that, when loaded into a computer, causes the computer to perform one or more functions described in the embodiments. The program may be stored on a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, the computer-readable medium or tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology. The computer-readable medium or tangible storage medium includes CD-ROM, digital versatile disc (DVD), Blu-ray disc or other optical disk storage. The computer-readable medium or tangible storage medium includes magnetic tape, magnetic disk storage or other magnetic storage devices. The program may be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, the transitory computer-readable medium or communication medium includes electrical, optical, acoustic or other forms of propagating signals.

Although the embodiments have been described above, the configurations of the above-mentioned embodiments may be combined with each other, or some of the configurations may be replaced with other configurations. Furthermore, the configurations of the above-mentioned embodiments may be modified in various ways without departing from the spirit of the invention.

Each drawing is merely an example for describing one or more embodiments. Each drawing may relate not only to one particular embodiment, but also to one or more other embodiments. As will be appreciated by those skilled in the art, various features or steps described with reference to any one drawing may be combined with features or steps shown in one or more other figures to create, for example, an embodiment not explicitly shown or described. Not all features or steps shown in any one drawing are necessary to describe an exemplary embodiment, and some features or steps may be omitted. The order of steps described in any drawing may be changed as appropriate.

REFERENCE SIGNS LIST 1 Caller terminal 10 Information processing system 100 Information processing device 110 Reception unit 120 Determination unit 130 Answer generation unit 131 Signal conversion unit 132 Prompt generation unit 133 Language model selection unit 134 Search function unit 140 Connection processing unit 150 Communication unit 160 Storage unit 161 Selection information 162 Customer information 200 Server 210 First language model 220 Second language model 300 Staff terminal 400 Reference information storage unit 1000 Computer 1010 Bus 1020 Processor 1030 Memory 1040 Storage device 1050 Input/output interface 1060 Network interface N1 First network N2 Second network P1 Caller P2 Staff

Claims (9)

a reception unit for receiving a method selection signal regarding a reply method from a caller terminal;
A determination unit that determines whether the sender is requesting a response by a generative artificial intelligence or a human based on the method selection signal;
an answer generation unit that, when it is determined that the caller requests the answer from the generative artificial intelligence, accepts an inquiry from the caller, selects one generative artificial intelligence from a plurality of generative artificial intelligences prepared in advance, causes the selected generative artificial intelligence to generate the answer, and transmits the answer generated by the generative artificial intelligence to the caller terminal;
a connection processing unit that connects the caller terminal and a staff terminal so that they can communicate with each other when it is determined that the caller has requested a human response ;
The answer generation unit selects one of the generative artificial intelligences based on the content of words included in the inquiry uttered by the sender and selection information linking the types of the generative artificial intelligences with a plurality of words expected to be included in the inquiry content.
Information processing device. 2. The information processing device according to claim 1,
The answer generation unit selects one of the generative artificial intelligences based on the method selection signal, which is a tone signal in a telephone function of the caller terminal or voice data uttered by the caller.
Information processing device. 2. The information processing device according to claim 1,
the answer generation unit transmits to the caller terminal voice data, which is data to be transmitted to the caller terminal via a first network, and text data, which is data to be transmitted to the caller terminal 1 via a second network different from the first network;
Information processing device. 2. The information processing device according to claim 1,
The answer generation unit transmits the answer generated by the generative artificial intelligence to the caller terminal as text data.
Information processing device. 2. The information processing device according to claim 1,
The answer generation unit converts the answer generated by the generative artificial intelligence into voice data and transmits it to the caller terminal, and then accepts an inquiry from the caller again in response to the caller's response.
Information processing device. In the information processing device according to any one of claims 1 to 5,
The answer generation unit converts the inquiry, which is an utterance of the caller, into text, and generates a prompt for causing one of the generative artificial intelligences to generate the answer based on the converted text.
Information processing device. The information processing device according to claim 6 further comprises:
the answer generation unit searches a predetermined database for information related to the inquiry, and generates the prompt using the search results.
Information processing device. The computer
receiving a method selection signal regarding the answering method from the caller terminal;
determining whether the caller is requesting a generative artificial intelligence response or a human response based on the method selection signal;
When it is determined that the caller has requested the answer from the generative artificial intelligence, the method accepts an inquiry from the caller, selects one generative artificial intelligence from a plurality of generative artificial intelligences prepared in advance, causes the selected generative artificial intelligence to generate the answer, and transmits the answer generated by the generative artificial intelligence to the caller terminal;
when it is determined that the caller has requested the answer by a person, the caller terminal and the staff terminal are connected so as to be able to communicate with each other;
1. An information processing method , comprising:
Selecting one of the generative artificial intelligences based on the content of words included in the inquiry, which is the utterance of the sender, and selection information linking a plurality of types of the generative artificial intelligences with a plurality of words expected to be included in the content of the inquiry.
Information processing methods. receiving a method selection signal regarding the answering method from the caller terminal;
determining whether the caller is requesting a generative artificial intelligence response or a human response based on the method selection signal;
When it is determined that the caller has requested the answer from the generative artificial intelligence, the method accepts an inquiry from the caller, selects one generative artificial intelligence from a plurality of generative artificial intelligences prepared in advance, causes the selected generative artificial intelligence to generate the answer, and transmits the answer generated by the generative artificial intelligence to the caller terminal;
when it is determined that the caller has requested the answer by a person, the caller terminal and the staff terminal are connected so as to be able to communicate with each other;
1. An information processing method, comprising:
Selecting one of the generative artificial intelligences based on the content of words included in the inquiry, which is the utterance of the sender, and selection information linking a plurality of types of the generative artificial intelligences with a plurality of words expected to be included in the content of the inquiry.
Making a computer execute an information processing method;
program.

Images