Content determination device, display control method, and computer program

Abstract

To provide a content determination device for checking a result of determination of whether it is an AI content or a non-AI content.SOLUTION: A display control unit displays, on a display window, a determination result obtained by determining whether a determination object content is an AI content generated by AI, or a non-AI content generated by means other than AI. On the display window, for example, a probability value that the determination object content is the AI content is displayed.SELECTED DRAWING: Figure 7

Description

[Technical field]

[0001] The present invention relates to a content determination device, a display control method, and a computer program. [Background technology]

[0002] There are two types of content: AI-generated content and non-AI-generated content. Image generation AI includes an artificial intelligence program called Midjourney (see Non-Patent Document 1). Text generation AI includes an artificial intelligence chatbot called ChatGPT (see Non-Patent Document 2 and Non-Patent Document 3). [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Midjourney Home Page, [Retrieved August 21, 2023], Internet<URL:https: / / www.midjourney.com / home / ?callbackUrl=%2Fapp%2F> [Non-Patent Document 2] ChatGPT Homepage, [Retrieved August 21, 2023], Internet<URL:https: / / chat.openai.com / auth / login> [Non-Patent Document 3] stable diffusion homepage, [searched on September 8, 2023], Internet<URL:https: / / ja.stability.ai / stable-diffusion> Summary of the Invention [Problem to be solved by the invention]

[0004] AI content is difficult to distinguish from non-AI content. For example, it is extremely difficult to distinguish whether the content provided on a web page is AI content or non-AI content. [Means for solving the problem]

[0005] A content determination device for solving the above problem includes a determination unit that determines whether the content to be determined is AI content generated by an AI or non-AI content generated by a means other than an AI, and a display control unit that displays the determination result of the determination unit on a display screen.

[0006] In addition, the display control method for solving the above problem is a computer program for executing the display control method, in which a display control unit displays a determination result as to whether a content to be determined is an AI content generated by an AI or a non-AI content generated by a means other than an AI, on a display screen.

[0007] According to the above configuration, the determination result as to whether the content to be judged is AI content or non-AI content is displayed on the screen, making it possible to easily confirm whether the content is AI content or non-AI content.

[0008] In the content determination device, the display control method, and the computer program, the display screen may include a determination target display area and a result display area, and the display control unit may be configured to display the determination target content in the determination target display area and display a binary classification determination result as the determination result in the result display area. According to the above configuration, the binary classification determination result can be displayed as the determination result on the display screen to inform the user.

[0009] In the content determination device, the display control method, and the computer program, the display screen may include a determination target display area and a result display area, and the display control unit may display the determination target content in the determination target display area and display a probability value that the determination target content is the AI ​​content in the result display area. According to the above configuration, the possibility that the determination target content is the AI ​​content can be notified to the user by displaying the probability value on the display screen.

[0010] In the content determination device, the display control method, and the computer program, the content to be determined may be image content, and the determination unit may determine, for each of a plurality of regions in the content to be determined, whether it is an AI region generated by an AI or a non-AI region generated by a method other than an AI. According to the above configuration, it is possible to determine, for each of a plurality of regions in the content to be determined, whether it is an AI region or a non-AI region.

[0011] The content determination device, the display control method, and the computer program may further include a classification selection unit that selects a classification of the content to be determined based on a classification designation command inputted by an operation unit and designating one classification from among a plurality of predetermined classifications, and the determination unit may determine whether the content to be determined associated with the classification designated in advance is the AI ​​content or the non-AI content. According to the above configuration, it is possible to determine whether the content to be determined is an AI content or a non-AI content after designating gender, etc., of the content to be determined in advance.

[0012] The above computer program can be installed in a computer to realize a content determination device. The above computer program is distributed through a network or through a portable recording medium such as an optical disk or a semiconductor memory, and is installed in the computer. Effect of the Invention

[0013] According to the present invention, the determination result of whether the content is AI content or non-AI content can be easily confirmed. [Brief description of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram showing a configuration of a content determination system according to an embodiment. [Diagram 2] FIG. 2 is a block diagram of a terminal device in the embodiment. [Diagram 3] FIG. 3 is a block diagram of a content determination device in the embodiment. [Figure 4] FIG. 4 is a diagram illustrating a determination function of the content determination device in the embodiment. [Diagram 5] FIG. 5 is a diagram showing an upload start display screen for uploading the determination target content in the embodiment. [Figure 6] FIG. 6 is a diagram showing a judgment start display screen for starting judgment of judgment target contents in the embodiment. [Figure 7] FIG. 7 is a diagram showing a judgment result display screen showing the judgment result of the judgment target content in the embodiment. [Figure 8] FIG. 8 is a diagram showing a determination result display screen showing the determination result of the determination target content in terms of an AI generation rate in the embodiment. [Figure 9] FIG. 9 is a diagram showing a judgment result display screen that explains the judgment result of the judgment target content by using a mask shading and a graph in the embodiment. [Figure 10] FIG. 10 is a diagram showing a judgment result display screen that explains the judgment result of the judgment target content (moving image) in a graph in the embodiment. [Figure 11] FIG. 11 is a diagram showing a judgment start display screen on which a classification of judgment target contents is designated in advance by an operation in the embodiment. [Figure 12] FIG. 12 is a diagram for explaining the classification of face types. [Figure 13] FIG. 13 is a diagram showing a judgment result display screen which shows the judgment result of the judgment target content after the face type has been classified in advance. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] A content determination device, a display control method, and a computer program to which the present invention is applied will be described below with reference to the drawings. [First embodiment] [Overall structure] 1, the content determination system 1 includes a plurality of client terminals 10 and a content determination device 30 serving as a server accessed by the client terminals 10. The client terminals 10 and the content determination device 30 are connected to each other via a network 2 such as the Internet. The network 2 includes the Internet, a Wide Area Network (WAN), a Local Area Network (LAN), a provider terminal, a wireless communication network, a wireless base station, a dedicated line, and the like. The client terminals 10 are connected to the content determination device 30 by wire or wirelessly.

[0016] The client terminal 10 is a portable or small information processing terminal such as a smartphone, tablet terminal, or mobile phone carried by a pedestrian. It is also an information processing terminal such as a desktop personal computer or a laptop computer. Furthermore, a headset or goggles using XR (Extended Reality / Cross Reality) technology such as VR (Virtual Reality), AR (Augmented Reality), or MR (Mixed Reality) may be connected to these information processing terminals. Furthermore, a smart watch or a smart speaker may be connected to the information processing terminal. Furthermore, the headset or goggles, smart watch, smart speaker, etc. may be directly connected to the network 2 as an information processing terminal.

[0017] The client terminal 10 can acquire various contents by downloading them through the network 2 or from other external storage devices. Here, the contents include images, videos, documents (text), music, audio, etc. The contents acquired by the client terminal 10 include AI contents generated by AI (Artificial Intelligence) and non-AI contents generated by humans. A user who acquires content on the client terminal 10 may acquire the content recognizing that it is AI content, or may acquire the content without recognizing whether it is AI content. Content acquired without recognizing whether it is AI content is unknown content, and it is unclear whether it is AI content.

[0018] A user can determine whether the content that the user has stored in the client terminal 10 is AI content or non-AI content by transmitting the content to the content determination device 30 as content to be determined.

[0019] The content determination device 30 determines whether the content to be determined transmitted from the client terminal 10 is AI content or non-AI content. Then, the content determination device 30 displays the determination result on a display screen such as a monitor of the client terminal 10.

[0020] [Client terminal] As shown in FIG. 2, the client terminal 10 includes a camera 11, a microphone 12, a speaker 13, a touch panel 14, a communication unit 15, a memory 16, and a control unit 17.

[0021] The camera 11 is an image input unit. The camera 11 is a CMOS camera, a CCD camera, or the like. The camera 11 may be an element built into the case of the client terminal 10, or may be connected to the main body of the client terminal 10 by wire or wirelessly. The camera 11 can capture images and videos including a subject and a background. The microphone 12 is an audio input unit. The microphone 12 collects surrounding audio in synchronization with, for example, capturing a video. The speaker 13 is an electroacoustic conversion element. The speaker 13 may be a speaker built into the main body, or may be an earphone or a headset connected to the main body by wire or wirelessly.

[0022] The touch panel 14 is a display unit and an operation unit. The touch panel 14 has a screen on which images or videos being captured or played are displayed. The screen also displays icons for performing various operations. The icons include button icons such as an application startup icon, a recording start button, and an upload button. When an icon is touched, the touch panel 14 generates an operation signal corresponding to the icon.

[0023] The client terminal 10 may be connected to a keyboard, a mouse, etc. Also, the client terminal 10 may be connected to a display device such as a liquid crystal display panel, an organic EL panel, etc.

[0024] The communication unit 15 is a communication interface that performs wired or wireless communication through a wide area network (WAN), a local area network (LAN), a wireless communication network, etc. The communication unit 15 communicates with the content determination device 30 and other servers. Note that the communication unit 15 may be connected to a headset or goggles, or may be connected to a smart watch, a smart speaker, etc.

[0025] The memory 16 includes a non-volatile memory, a hard disk, etc. The memory 16 holds application programs. The application programs include an operating system and a browser 16a for displaying web pages. In addition, there are image creation and editing application programs for creating and editing images and videos as content. There is also a text creation and editing program for creating documents.

[0026] The memory 16 is also provided with a data storage unit 16b. The data storage unit 16b stores content. Specifically, it stores image and video data captured by the camera 11. It also stores audio data collected by the microphone 12. It also stores content acquired from a web page. The content may be AI content, non-AI content, or unknown content.

[0027] For example, in the case of images or videos, AI content and non-AI content can be obtained by downloading from a webpage. Also, AI content is AI image content created using an image generation AI. Furthermore, it is content that combines AI image content and non-AI image content, for example, composite image content that combines a background non-AI image with a person AI image. Also, it is composite image content that combines a background AI image with a person non-AI image.

[0028] In the case of documents, AI text content and non-AI text content can be obtained by downloading them from web pages. AI text content created using text generation AI. Furthermore, composite text content that combines AI text content and non-AI text content.

[0029] Furthermore, in the case of audio, both AI audio content and non-AI audio content can be obtained by downloading from a webpage. AI audio content created using voice generation AI. Furthermore, composite audio content that combines AI audio content and non-AI audio content. Audio content includes book and newspaper reading content, music content, etc. Video content is a combination of video content and audio content, such as movie content.

[0030] The control unit 17 is, for example, a CPU. The control unit 17 starts an application program and controls the operation according to an input from the operation unit. The control unit 17 can start the browser 16a to browse a web page. For example, the control unit 17 accesses the content determination device 30 and uploads the determination target content, which is unknown content. Then, the control unit 17 can cause the content determination device 30 to determine whether the determination target content is AI content or non-AI content, and display the determination result on the display screen.

[0031] [Content Determination Device] 3, the content determination device 30 includes a communication unit 31, a storage unit 32, a control unit 33, and a display control unit 34. The communication unit 31 communicates with a number of information processing devices including the client terminal 10 through the network 2.

[0032] 3 and 4, the storage unit 32 is composed of a hard disk drive, a solid state drive (SSD), etc., and stores various programs and data. The storage unit 32 stores various programs, learning models, etc. for functioning as the determination unit 37. The display control unit 34 generates display data to be displayed on the display screen of the client terminal 10.

[0033] The judgment unit 37 includes a judgment program 35 and a learning model 36. The judgment program 35 is a program that uses the learning model 36 to judge whether the content to be judged is AI content or non-AI content.

[0034] The learning model 36 is a model created by deep learning. The learning model 36 is a learning model for determining whether the content to be determined is AI content or non-AI content.

[0035] In the case of image content, when creating the learning model 36, AI image content created by AI and non-AI image content created by humans are acquired by crawling. Next, the AI ​​image content and non-AI image content are input to the input layer. Then, a data model is created in which the feature points are rewritten as AI data and non-AI data. The learned learning data is saved in, for example, an HDF5 (Hierarchical Data Format 5) file format, and is used as the learning model 36.

[0036] The judgment unit 37 configured as above executes the judgment program 35, and judges whether the feature points of the judgment target content are closer to the AI ​​feature data or the non-AI feature data, or to what extent they match, according to the learning model 36, and generates a score. The score is, for example, a continuous value (for example, 0 to 1.0), and is a probability value that increases as the possibility of AI increases. Furthermore, a score of, for example, 0.5 is set as a threshold, and the judgment unit 37 judges the content to be non-AI content when the score is 0.5 or less, and judges the content to be AI content when the score is greater than 0.5. The threshold is not limited to 0.5. The score is displayed as an AI generation rate.

[0037] Before performing the determination process, the determination unit 37 performs image recognition processing on the content to be determined, thereby distinguishing between a person region and a background region, etc. The determination unit 37 distinguishes a face region from a person region. The determination unit 37 distinguishes the face region into a hair region, an eye region, a nose region, etc. The determination unit 37 analyzes, for example, the face region and uses facial features resulting from the analysis for face determination.

[0038] When creating the learning model 36, in the case of a person, a learning model may be created for each gender (male and female), etc. The determination unit 37 selects one of the learning models prepared for each gender according to a classification designation command that designates the gender. In this way, by preparing a learning model for each classification such as gender, it is possible to improve the determination accuracy. Also, a learning model may be created for each of the hair region, eye region, nose region, etc. in the face region.

[0039] Also, a learning model may be created for each face type described below. In this case, the determination unit 37 performs image recognition processing on the content to be determined, analyzes the face area, and determines the face type. Then, the determination unit 37 selects a learning model for the determined face type from the learning models created for each face type. In this way, by preparing a learning model for each face type, the determination accuracy can be improved.

[0040] The display control unit 34 generates display data to be displayed on the display screen of the client terminal 10 according to the judgment result. For example, in the case of a binary classification judgment, the display control unit 34 generates display data for displaying that the image content is AI image content when the score output from the output layer is greater than a threshold value. Also, the display control unit 34 generates display data for displaying that the image content is non-AI image content when the score output from the output layer is less than a threshold value.

[0041] When the score (probability value) is displayed as an AI generation rate, the display control unit 34 generates display data for displaying the score as an AI generation rate in %, and further generates display data for displaying the determination result of whether each of the background region, hair region, eye region, and nose region in the person region is an AI region or a non-AI region as a binary value of AI or non-AI, or as an AI generation rate.

[0042] [Content Judgment Processing] Next, the user accesses the content determination device 30 with the client terminal 10, and uploads the unknown image content stored in the data storage unit 16b by the user to the content determination device 30 as the content to be determined. FIG. 5 shows an upload start display screen displayed on the client terminal 10. The display control unit 34 displays the display screen on the client terminal 10. The user uses the operation unit of the client terminal 10 to drag and drop the file 41 of the content to be determined stored in the data storage unit 16b into the upload area 42 displayed on the display screen. The user selects the file 41 of the content to be determined stored in the client terminal 10, and touches the upload button for uploading to perform an operation to upload.

[0043] 6 shows a judgment start display screen when AI judgment is started when uploading a file 41 of the judgment target content to the content judgment device 30 is completed, and the display control unit 34 displays this display screen on the client terminal 10. The judgment start display screen is provided with a judgment target display area 43 in which the uploaded judgment target content is displayed. Adjacent to the judgment target display area 43 are a result display area 44 in which the judgment result is displayed, a judgment button 45 for starting the judgment, and a comment area 46 in which comments regarding the judgment result are displayed. When the judgment button 45 is touched, the client terminal 10 transmits a judgment start command to the content judgment device 30.

[0044] In addition, on the judgment start display screen, a program that automatically uploads the content to be judged may be started in the background, the judgment button 45 may be omitted, and the judgment process may be automatically started as soon as the upload is completed.

[0045] FIG. 7 is a judgment result display screen showing the judgment result. The display control unit 34 displays the display screen on the client terminal 10. The content judgment device 30 judges the judgment target content in accordance with the judgment start command. The judgment unit 37 judges the judgment target content as AI content when the score of the judgment target content is higher than the threshold. The judgment unit 37 judges the judgment target content as non-AI content when the score is lower than the threshold. Then, when the judgment process is completed, the content judgment device 30 displays "AI" in the result display area 44 when the judgment target content is AI content (see FIG. 7). Furthermore, the content judgment device 30 displays "non-AI" when the judgment target content is non-AI content.

[0046] In the case of a binary classification judgment, for example, when the content is AI content, the word "AI" may be displayed in a relatively darker color and in bold type. When the content is non-AI content, the word "non-AI" may be displayed in a relatively lighter color and in thin type. When the content is AI content, a first color such as red may be used, and when the content is non-AI content, a second color such as black may be used. When the content is AI content, a robot-like object may be displayed, and when the content is non-AI content, a human object may be displayed. Furthermore, when the content is AI content, "AI" may be displayed, and when the content is non-AI content, nothing may be displayed.

[0047] In the content determination device 30, the determination unit 37 distinguishes between the background region, the hair region of the person region, the eye region, the nose region, etc., by image recognition processing. The determination unit 37 specifies the region in which the feature points of the content to be determined that are far from the AI ​​data are located, and based on this, the display control unit 34 generates a comment to be displayed in the comment area 46. In Fig. 7, the display control unit 34 displays a comment indicating the reason for the determination in the comment area 46, such as "It is highly likely that the area around the eyes was generated by AI."

[0048] [Effects of the first embodiment] The first embodiment as described above can provide the following advantages. (1-1) The judgment result display screen displays the judgment result of whether the content to be judged is AI content or non-AI content. This allows the user to easily check whether it is AI content or non-AI content. In particular, since the binary classification judgment result is displayed, the user can clearly know the judgment result.

[0049] (1-2) The reason why the content to be judged is determined to be AI content is displayed in the comment area 46. This makes it easier for the user to understand the reason why the content to be judged is AI content or non-AI content.

[0050] Second Embodiment The second embodiment is another aspect of the judgment result display screen. In Fig. 8, the AI ​​generation rate is displayed in the result display area 44. Here, the judgment unit 37 displays the possibility (%) that the content is AI content according to the score. In the result display area 44 in Fig. 8, 80% is displayed.

[0051] For example, when the result display area 44 displays the possibility that the content to be judged is an AI content as a probability value (%), the higher the AI ​​probability, the darker the color or the thicker the text may be. Also, the lower the AI ​​probability, the lighter the color or the thinner the text may be.

[0052] [Effects of the second embodiment] The second embodiment as described above can provide the following advantages. (2-1) On the judgment result display screen, the possibility that the judgment target content is AI content can be displayed as a probability value (%) to inform the user. For example, when a value of around 50% is displayed in the result display area 44, the user can understand that it is difficult to distinguish whether the content is AI content or non-AI content.

[0053] Third Embodiment The third embodiment is another aspect of the judgment result display screen. In FIG. 9, a mask 47 is superimposed on the image of the judgment target content displayed in the judgment target display area 43, like a thermograph, and the areas with high and low AI generation rates are indicated by the shading of the mask 47. The higher the AI ​​generation rate, the higher the density (lower the transmittance) of the mask 47 displayed. The lower the AI ​​generation rate of the mask 47, the lower the density (higher the transmittance) of the mask 47 displayed. In the judgment target display area 43 in FIG. 9, the mask 47 is dark in the area from the right eye to the right cheek and mouth of the person, indicating that this is an area with a relatively high AI generation rate. The area around the dark area is light, indicating that this is an area with a relatively low AI generation rate.

[0054] The AI ​​generation rate is also displayed in graph 48. In the example of Fig. 9, it is displayed as a bar graph. The AI ​​generation rate in each of the hair region, eye region, nose region, and background region is displayed as a bar graph. The graph display can also be displayed as a line graph or the like.

[0055] [Effects of the third embodiment] The third embodiment as described above can provide the following advantages. (3-1) The user can check the AI ​​generation rate in any area of ​​the image of the content to be judged by checking the shading of the mask 47.

[0056] (3-2) The AI ​​generation rate of each region can be easily visually recognized by graph 48. The AI ​​generation rate can be compared for each region by two elements: the shading by mask 47 and graph 48.

[0057] [Fourth embodiment] The fourth embodiment shows a form of a judgment result display screen in the case where the judgment target content is a video content that is a collection of images. In Fig. 10, frame images that make up the video are displayed every few frames in the judgment target display area 43. Then, judgment is made for each frame image that makes up the video (it may be frame images of all frames, or may be frame images every few frames) in the same way as for the image content.

[0058] A graph 51 is provided below the judgment target display area 43. In graph 51, the horizontal axis is the time axis, and the vertical axis indicates the AI ​​generation rate. In graph 51, the change in the AI ​​generation rate over time can be confirmed. In graph 51, the horizontal axis indicates the average value of the AI ​​generation rate of each frame image judged over one second (80%, 85%, 90%, 85% in graph 51).

[0059] The result display area 44 shows the average AI generation rate of all frames for which judgment has been performed, while the result detail display area 52 shows the AI ​​generation rate of each frame image.

[0060] [Effects of the fourth embodiment] The fourth embodiment as described above can provide the following advantages. (4-1) It is possible to judge whether not only image content but also video content is AI content or non-AI content. Furthermore, on the judgment result display screen, you can also know the AI ​​generation rate for each frame image that makes up the video.

[0061] Fifth embodiment The fifth embodiment is another aspect of the judgment start display screen. In FIG. 11, the judgment target display area 43 displays the judgment target content. A classification selection area 56 in which the user specifies the classification of the judgment target content is provided adjacent to the judgment target display area 43. A first classification selection section 57 is provided at the upper stage to select the target for which judgment is performed as to whether it is AI content or non-AI content. The first classification selection section 57 includes a person button 57a, a landscape button 57b, and an animal button 57c. The first classification selection section 57 may be configured to select one classification from among multiple classifications in a pull-down menu. In this embodiment, the person button 57a ​​is selected to perform judgment on a person in the judgment target content displayed in the judgment target display area 43.

[0062] Further, a second classification selection section 58 is provided below the first classification selection section 57. The second classification selection section 58 can select a sub-classification of the classification selected in the first classification selection section 57. In this embodiment, the first classification selection section 57 selects "person". Therefore, the second classification selection section 58 allows the selection of gender (male or female). In this embodiment, female is selected.

[0063] In the present embodiment, the learning model includes a learning model according to the classification of the first classification selection unit 57. In the present embodiment, the learning models for people include a learning model for men and a learning model for women. In the present embodiment, when "person" is selected, the determination unit 37 selects the learning model for women. The determination unit 37 uses the learning model for women to determine whether the content to be determined is AI content or non-AI content.

[0064] In addition, when a landscape is selected in the first classification selection unit 57, the second classification selection unit 58 can select, for example, a daytime scene or a nighttime scene. In such a case, a learning model for a daytime scene and a learning model for a nighttime scene are prepared as learning models. In addition, when an animal is selected in the first classification selection unit 57, the second classification selection unit 58 can select, for example, a dog, a cat scene, a bird, etc. In such a case, a learning model for a dog, a learning model for a cat, a learning model for a bird, etc. In addition, the classification in the first classification selection unit 57 is not limited to "people", "scenery", and "animals".

[0065] 11 may be divided into a background region and a person region. In the classification selection area 56, it may be possible to select whether to make the person region or the background region the object of judgment.

[0066] [Effects of the fifth embodiment] The fifth embodiment as described above can provide the following advantages. (5-1) In the first classification selection unit 57 and the second classification selection unit 58, a learning model according to the specification can be selected by the user designating the classification of the content to be judged in advance. This makes it possible to more accurately determine whether the content to be judged is AI content or non-AI content.

[0067] (5-2) The second classification selection section 58, which is a sub-classification section, is provided below the first classification selection section 57. This allows the classification of the content to be determined to be specified in more detail with a simple operation.

[0068] Sixth Embodiment The sixth embodiment is an example in which a face type is determined in advance in addition to the fifth embodiment (see FIG. 11). Face types are as shown in FIG. 12. A person's face type can be classified into any of the following on a coordinate plane: a first quadrant of child×curve, a second quadrant of child×straight line, a third quadrant of adult×straight line, and a fourth quadrant of adult×curve. When a determination button 45 is touched on the display screen shown in FIG. 13, for example, the determination unit 37 determines which category the content to be determined belongs to in accordance with a determination start command.

[0069] FIG. 13 shows a judgment result display screen in the sixth embodiment. The judgment result display screen is provided with a first face type judgment result display area 61. The first face type judgment result display area 61 displays a coordinate plane. The judgment unit 37 analyzes the face area to judge the face type. For example, the judgment unit 37 can analyze the face area of ​​the judgment target content to judge that it corresponds to "active cute" included in the first quadrant and specify the type. For example, the judgment unit 37 has a learning model for performing face judgment, and can also judge which type the person in the judgment target content is. In the example of FIG. 13, the person in the judgment target content is judged to be "active cute", and the display control unit 34 displays the judgment target content in the first quadrant on the coordinate plane. In addition, a second face type judgment result display area 62 is provided to display the result of the face judgment. In this embodiment, "active cute" is displayed.

[0070] The face type may be determined by having the user answer a number of questions. In this case, a number of classification selection units may be provided, such as a first classification selection unit 57, a second classification selection unit 58, a third classification selection unit, etc., and the user may answer the questions. For example, the first classification selection unit 57 may prompt the user to answer "yes / no" to a question such as "your face is on the long side," and the second classification selection unit 58 may prompt the user to answer "yes / no" to a question such as "your face is rather childlike."

[0071] In the sixth embodiment, in parallel with the face determination, as in the fifth embodiment, in the classification selection area 56, "person" is selected in the first classification selection unit 57, and "woman" is selected in the second classification selection unit 58. The determination unit 37 selects a learning model for women. Then, the determination unit 37 uses the learning model for women to determine whether the content to be determined is AI content or non-AI content.

[0072] The sixth embodiment as described above can provide the following advantages. (6-1) In the first classification selection unit 57 and the second classification selection unit 58, the user can specify the classification of the content to be judged in advance, and a learning model according to the specification can be selected. This makes it possible to more accurately determine whether the content to be judged is AI content or non-AI content. In addition, the result of face judgment of the person in the content to be judged can also be known.

[0073] In addition, when performing face determination, a learning model corresponding to the first to fourth quadrants on the coordinate plane may be created. With this configuration, the classification of the content to be determined can be specified in advance by face determination. The determination unit 37 can select a learning model according to the classification according to this classification specification command. This makes it possible to more accurately determine whether the content to be determined is AI content or non-AI content.

[0074] [Other Modifications] The first to sixth embodiments described above can be further modified as appropriate as follows: The above-described embodiments and the following modifications can be combined with each other to the extent that there is no technical contradiction.

[0075] In the case of audio, AI audio content created using an audio generation AI and non-AI audio content created by humans, etc., can be obtained by crawling, and feature points can be obtained from each audio content to create a learning model. In addition, when making a judgment, it is possible to specify a classification for the content to be judged, such as gender. In this case, a learning model is prepared for each classification. Then, by configuring the learning model to be selected according to the classification specification, the judgment accuracy can be improved. As a classification, a learning model of animal cries may be prepared in addition to gender. Furthermore, when judging music, a learning model may be prepared according to classifications such as pop music, jazz music, opera music, classical music, etc.

[0076] In the case of text, AI text content created using a text generation AI and non-AI text content created by humans, etc. can be obtained by crawling, and feature points can be obtained from each text content to create a learning model. In addition, when making a judgment, it is possible to specify a classification for the content to be judged, such as papers, novels, reviews, fiction, history, geography, news, etc. In this case, a learning model is prepared for each classification. Then, by configuring the system to select a learning model according to the classification specification, the judgment accuracy can be improved.

[0077] When displaying the possibility of AI content as a probability in the result display area 44, 0% to 100% may be divided into a high possibility of AI (e.g., 70% to 100%), a low possibility of AI (e.g., 0% to 30%), and an unknown possibility of AI (e.g., 31% to 69%). When the possibility of AI is high, a first object such as "◯", "robot", or "strong" may be displayed. When the possibility of AI is low, a second object such as "×", "person", or "weak" may be displayed. When it is unknown, a third object such as "△", "?", or "animal such as cat" may be displayed. The range of division is not limited to three, and may be two, four, or more. In such a case, a symbol may be associated with each range, and the symbol may be displayed.

[0078] The content determination device 30 may be configured to be able to directly input the content to be determined without going through the network 2. In this case, the content determination device 30 may be configured to display the determination result on a display device included in the device.

[0079] When a headset or goggles is an information processing terminal, the information processing terminal may be equipped with a judgment program 35 and a learning model 36 for configuring a judgment unit 37. In such a configuration, it is not necessary to upload the content to be judged to a server. In this case, after starting the judgment program in the information processing terminal, a judgment start display screen may be displayed on the display screen of the browser 16a, and each time an image content is displayed on the display screen, judgment may be performed on the image content and the judgment result may be displayed. Also, in the setting screen, the judgment program may be started in the background, the judgment result display screen may be omitted, and only the judgment result (for example, the result display area 44) may be popped up near the judged image judgment content. [Explanation of symbols]

[0080] 1. Content Evaluation System 2. Network 10. Client terminal 17...Control section 30...Content judgment device 35...Judgment Program 36…Learning Model 37...Judgment section 41...File 42...Upload area 43…Judgment target display area 44…Result display area 45…Judgment button 46...Comments area 47...Mask 48...Graph 51...Graph 52…Result details display area 56...Classification selection area 57…First classification selection section 58...Second classification selection section 61…First face type determination result display area 62…2nd face type determination result display area

Claims

1. A determination unit that determines whether the content to be determined is an AI content generated by an AI or a non-AI content generated by a means other than an AI; A display control unit that displays the determination result by the determination unit on a display screen. Content determination device.

2. the display screen includes a judgment target display area and a result display area; The display control unit displays the content to be judged in the judgment target display area, and displays a binary classification judgment result as the judgment result in the result display area. The content determination device according to claim 1 .

3. the display screen includes a judgment target display area and a result display area; The display control unit displays the judgment target content in the judgment target display area, and displays a probability value that the judgment target content is the AI ​​content in the result display area. The content determination device according to claim 1 .

4. the content to be determined is an image content, The determination unit determines, for each of a plurality of areas in the content to be determined, whether the area is an AI area generated by AI or a non-AI area generated by a means other than AI. The content determination device according to claim 1 .

5. The apparatus further includes a classification selection unit that selects a classification of the content to be determined based on a classification designation command that is input via an operation unit and designates one of a plurality of predetermined classifications, The determination unit determines whether the content to be determined, which is associated with a classification specified in advance, is the AI ​​content or the non-AI content. The content determination device according to claim 1 .

6. The display control unit displays the result of the determination as to whether the content to be determined is AI content generated by AI or non-AI content generated by a means other than AI on the display screen. Display control method.

7. A computer program that executes a display control method in which a display control unit displays on a display screen the result of a determination as to whether the content to be determined is AI content generated by AI or non-AI content generated by a means other than AI.

Images