Method for automatically implementing design guidelines, system for same, computer program product, and computer-readable storage medium

Abstract

The invention relates to a method for automatically implementing design guidelines, to a system for carrying out the method, to a computer program for same, and to a computer-readable storage medium. The invention allows, for the first time, the use of a generative AI for generating digital communication means for a specific institution or a business with specified design guidelines. In particular, the invention even allows the implementation of design guidelines, which correspond to regionally different cultures, using prompts so that an individual input for a series of prompts for generating a digital product presentation leads to regionally different results, all over the world for example, thereby reflecting in particular the respective customs and prevalent culture according to the prompt.

Description

[0001] Description

[0002] Method for the automated implementation of design guidelines, system therefor, as well as computer program product, and computer-readable storage medium

[0003] The invention relates to a method for the automated implementation of design guidelines and a system for carrying out the method, a computer program therefor and a computer-readable storage medium.

[0004] In human-machine interaction, a lot of working time, both for the machine and the human, is still lost because the interaction does not run smoothly.

[0005] For example, human users misunderstand the machine's outputs, the machine responds with an unmanageable number of possible results, and the user receiving the output is overwhelmed and repelled. There is therefore a need to optimize human-machine interaction, particularly for processes and systems operating over the Internet ("loT"), for example, in different countries or even worldwide. This ultimately not only improves user acceptance but also results in energy savings, quite apart from avoiding serious errors that can arise from misunderstood human-machine communication in process engineering, for example, in the control and regulation of processes in general.Errors in human-machine interaction can arise not only from unclear inputs and outputs but also from texts, gestures and other inputs and / or outputs that are misleading or even have completely different meanings in other cultures.

[0006] A style guide or design guideline describes how certain elements of a printed product or a website are to be designed and could result in adaptation to the respective cultural, social and / or industry circles.

[0007] Design guidelines can ensure a uniform appearance for an institution or company's various communication media, such as advertising and information material, product presentations, business letters or websites, and thus enable the creation of a corporate identity. Style guides are also used for software development or products in general; here too, the aim is for the products to be perceived by the public, particularly by international, cross-sector and / or all social classes, as belonging together and coming from a specific company, and also to be associated with, for example, a corporate brand.

[0008] In printed communication media, the design elements are expressed in - e.g. paper type, paper size, font, font color, printing techniques, finishes, line spacing, use of design elements, use of colors, logos but also the language, uniform designations, language style and formulations can be used as design elements in this sense.

[0009] In digital media, especially in web design, style guides are gaining importance in the design process. This is due to the increasing relevance of responsive web design (RWD), in which the individual, pixel-perfect layout is becoming less important.

[0010] The RWD is a design and technical paradigm for creating websites so that they can respond to the characteristics of the device used, especially smartphones and tablet computers.

[0011] The graphic structure of a "responsive" website is based on the requirements of the respective device with which the website is viewed. These particularly concern the arrangement and presentation of individual elements, such as navigation, page columns and text, but also the use of different input methods such as mouse with "click, hover" or touchscreen with "tap and swipe". The technical basis for the common web standards HTML5, CSS3 and Java Script.

[0012] HTML5 is the fifth version of the Hypertext Markup Language, a computer language for marking up and linking texts and other contents of electronic documents, primarily on the World Wide Web.

[0013] CSS stands for Cascading Style Sheets and, along with HTML and JavaScript, is one of the core languages ​​of the World Wide Web. CSS creates style instructions that are primarily used in conjunction with markup languages ​​such as HTML and XML.

[0014] XML stands for Extensible Markup Language and is a markup language for representing hierarchically structured data in the form of a text file that is readable by both humans and machines.

[0015] The disadvantage of these well-known tools for adapting digital content to all types of devices is that there is currently no tool available with which digital content can be automatically adapted, i.e. adapted to the design guidelines of a company, an institution and / or a cultural group.

[0016] Rather, this task still falls to human users, who either select the usable results that comply with design guidelines from a flood of results based on an input given to an AI, or - which is the state of the art - completely forgo the use and support of AI to generate digital communication media that comply with certain design guidelines. Adaptations to design guidelines for digitally generated communication media are therefore still carried out manually. This means that if an AI is used, it generates, for example, 500 representations of a product and a user then selects the appropriate one - after hours of detailed work, e.g. manually. The disadvantage of this is not only the unnecessary consumption of working time, but also the unnecessary consumption of computing capacity, energy and storage space.There is a need to overcome this technical disadvantage of conventional technology .

[0017] The previously common technique of "prompting", i.e. the method in connection with AI and machine learning which initiates one or more results, one or more responses, outputs and / or actions from an instruction signal, an input and / or a start of an AI, does not take into account soft skills and / or cultural, personal, identity-promoting specifications and / or competencies. This is disadvantageous, however, because it inhibits the spontaneity and / or mutual understanding with which the user and the machine interact. In the best case, this can result in a waste of time and / or energy, but in the worst case it can lead to misjudgments and incorrect results.

[0018] Large Language Model (LLM) technology enables the generation of images, generating an inflated number of uncontrolled, irregularly styled and / or designed results. Currently, there is no style guide or design guidelines that automatically regulate this flood of images and / or content.

[0019] Therefore, it is the object of the present invention to automatically implement design guidelines and / or "style guides" in LLM-generated content, such as images, information, texts. The aim is also to promote clarity in communication in human-machine interaction and thereby to optimize effectiveness, user acceptance, energy consumption, working time, consumption of computing capacity, energy, storage space and / or computer usage, because the implementation of design guidelines early in the processing automatically reduces the number of generated results and thus saves computing time.A further object of the invention is to provide a method and a system as a digital tool with the aid of which design guidelines are automatically implemented in inputs for an AI and / or content generated by AI, i.e. results and outputs, so that the number of surplus, superfluous and therefore deleted results, which uselessly consume energy and / or computing time of the AI, is kept as low as possible.

[0020] This object is achieved by the subject matter of the present invention as disclosed in the description, the claims and the figure.

[0021] Accordingly, the subject of the present invention is a method for the automated implementation of design guidelines in the output of a result generated with the aid of generative artificial intelligence "AI", comprising the following method steps:

[0022] - Digital recording of the circumstances of the input and output device,

[0023] - Selection of appropriate design guidelines and

[0024] - automated implementation of the appropriate design guidelines into the process from user input through AI calculation to machine output,

[0025] Implementation of the design guidelines by defining the possible selection palette of one or more prompts and / or by generating digital boundaries for the generative AI to compare with the contents and / or results calculated by the AI, storing the generated digital boundaries within the prompt and / or within the processor(s) on which the AI ​​is running,

[0026] - Input of one or more tasks by a user to the generative AI

[0027] - Calculating the possible outcomes that lie within these digital boundaries using generative AI

[0028] - Output of the results taking into account the selected design guidelines, whereby

[0029] - the automatic selection of the results by means of the digital boundaries and / or design guidelines takes place either during input via the prompt, during calculation by the AI ​​and / or during output by automatically comparing the results generated by the generative AI with the specified digital boundaries and / or with the specified digitally available design guidelines within the AI.

[0030] Furthermore, the subject of the present invention is a system for carrying out the method described above, wherein the system has an interface which has means for recording the circumstances and for the automated selection and implementation of design guidelines, as well as prompts for requesting and / or inputting and forwarding to the generative AI, wherein the interface between the user and the processor on which the AI ​​is running is provided in such a way that this interface is coupled to the prompt, so that the prompt is set up and configured by means of the interface in such a way that it only forwards inputs which are within predetermined digital boundaries to the generative AI, wherein variants and possibilities from the input which are not within the digital boundaries are identified and removed and / or deleted by the prompt.Furthermore, the invention relates to a computer program product with program code means which cause an electronic computing device, such as one or more processors which are configured and arranged to receive inputs, to process them and to forward them to a generative AI for execution, to carry out a method according to one of the preceding claims when the program code means are processed by the electronic computing device.

[0031] Finally, the invention relates to a computer-readable storage medium with a computer program product as described above.

[0032] Such a process not only saves valuable computing time for the AI ​​and the system used to create the 500 variants, but also valuable time and energy for a corresponding output device, a monitor or printer, and paper. Last but not least, it saves the time of a human editor who, with his or her diverse range of perceptions, can intuitively distinguish between a representation that is “correct” and has a positive effect in the sense of the design guidelines, e.g. the “corporate identity”, and one that may be rated equally well by the AI ​​but which a human would not perceive as being so positive, not as clearly belonging to the company or culture, and / or may even find it repulsive.

[0033] While the AI ​​results can also be selected by comparing the possible results with the digital boundaries stored in the generative AI, this is preferably done by correspondingly restricting the input options via the system prompt. This means that certain features that would otherwise be available to the system are automatically no longer available due to the specifications that the user has fundamentally set as digital boundaries for the system prompt in accordance with the design guidelines. The system's capabilities are thus limited to those generative AI results that lie within the digital boundaries that the design guidelines reflect.The design guidelines can also influ ence the optimization of a generative AI, so that the design guidelines adapt the input of a user for the AI ​​and / or the output of the AI ​​result to the skills, the culture, the background of the user collecting the result.

[0034] For example, a system used worldwide according to one embodiment of the invention can, by recording the circumstances, for example the cultural affiliation of the IP address of the output device, immediately "transfer" - i.e., use prompts, queries and / or input masks unchanged but adapted to the other cultural area.

[0035] A suitable interface for implementing the digital boundaries established by the design guidelines between the program that transmits user input to the generative AI is preferably one or more prompts. This seems most effective when the digital boundaries limit the pool of possibilities from the outset, similar to how the laws of nature impose digital limits on the possibilities generated by the AI ​​when solving physical problems, so that results that do not obey the laws of nature are not generated and made available by the AI ​​in the first place.

[0036] However, this does not preclude other interfaces, e.g. using filters, from making a selection from the generated results and / or outputs that are within the design guidelines as they were digitally captured.

[0037] When given input to the generative AI via a prompt, it can implement the digital boundaries in the input by removing, discarding and / or deleting the input choices that do not comply with the design guidelines.

[0038] A "prompt" is an input mask - for example a question and answer game on a monitor - with the help of which the user makes it clear to the generative AI what kind of question he has, what answers he wants and / or what kind of results he is looking for. The prompt is used, for example, to make the user's request understandable to the generative AI, to translate it and to transmit it so that the generative AI can calculate the corresponding results and return them to the user as a solution, result and / or answer.

[0039] Prompts are known from program input methods, in which the user is prompted to provide specific information as input to the system by placing specific labels at the front. This prefixing of specific labels depends on the programming language, operating system, program, whether the system is text-, visual-, and / or audio-based, and so on. This interface between artificial intelligence and the human user is generally called a "prompt."

[0040] The prompts also serve, for example, to clarify corporate affiliation and / or to enable the system to distinguish between system output and user input. An input follows the prompt, which remains attached to the input, so that the system clearly recognizes that it is input, not output.

[0041] An input can be made, for example, via mouse, keyboard, microphone, touchscreen, gestures, or light barrier. The prompt then responds via the same communication channel. The processing of inputs, for example, voice input via microphone received by an electronic device, is known, while input requests are provided in the form of prompts via a loudspeaker and / or a monitor. A user, for example, also has a sequence of prompts available while making a voice input.

[0042] This is particularly true when a significant number of functions or operations are accessible to a user, but the number of available functions or operations of the electronic device exceeds the number of inputs available via an input interface of the device, meaning the device cannot accept everything the user inputs. In this case, these requests, which are not immediately received by the device, are stored in a series of multiple prompts.

[0043] Every program, operating system, etc., has its own type of prompt input option, so the user knows from the prompt which program their input is being made in. A prompt refers not only to the identifier received from the processor before the input, but also to the identifier that terminates the input.

[0044] Prompting gives the user the opportunity to ask a question and / or a task within the system associated with the prompt, such as "design a picture of a pair of headphones." The system then answers the question and / or solves the task within the options provided by the system, including those provided via the prompt.

[0045] The general finding of the present invention is that by appropriate restrictions - "implementation of design guidelines" - in the form of digital boundaries, a number of results can be saved which, without these restrictions, would first be unnecessarily generated, then unnecessarily processed and executed, sent as output to corresponding end devices and finally discarded again through manual selection, thus unnecessarily consuming computing capacity, time, money and energy. A further aspect is that by adapting an input device to the human user via their local, cultural, contemporary, industry-standard and / or company-typical soft skills, unnecessary "talking past each other" can be avoided, i.e. the clarity and thus the effectiveness of a machine-human interaction is optimized. This brings advantages economically, in terms of energy policy and with regard to all aspects of sustainability.

[0046] It is advantageous if the digital boundaries are automatically implemented into the system through the "digital capture of the circumstances" and the selection or generation of appropriate design guidelines. Of course, establishing these digital boundaries can also increase efficiency because the computers and servers are used for only a fraction of the time.

[0047] In particular, it is also a general finding of the present invention that a globally oriented human-machine interaction should preferably adapt automatically to the respective cultural, personal and industry-specific practices, so that a uniform, "worldwide used software" is regionally, industry-specific and, if necessary, even equipped with a company identity.

[0048] In principle, it is always the same algorithm in the processor that is at work, but the input requests appear differently and are automatically designed differently depending on location, time of day, computer density in the room, temperature, time of year, etc. This will make the human user feel "understood" right from the start, because the machine does not request any inputs or deliver results that seem disturbing, strange, or alien. The implementation of individual restrictions in a generative AI directly and / or through a program input method using a prompt with predefined and defined digital boundaries that reflect design guidelines prevents generative AI from providing unusable solutions in the first place.As a result, the generative AI is not trained using a UN amount of results and manually selected data, but rather the generative AI is given digital limits in the form of boundary conditions, within which it only evaluates and forwards calculated solutions as results of a given task.

[0049] For example, LLM technology, or "Large Language Model Technology", enables the creation of images to display desired products, but so far it has produced a flood of uncontrolled, styled images.

[0050] Institutes and companies have therefore so far refrained from using generative AI to create solutions for digital communication media - for example slogans, texts, representations, images, websites - through which corporate identity is to be conveyed.

[0051] By introducing a prompt with digital boundaries for an LLM input, according to one embodiment of the method, a selection can already be made via the corresponding setting of the prompt. For example, the color palette of the available colors is clearly defined when entering the prompt - for example, by specifying a hex code. A corresponding LLM input could be: "Create a representation of an iPhone 13 centered on an azure blue background" and specifying the hex code for the azure blue. In the field of computer-aided design, a "hex code" refers to a system of hexadecimal color definitions, whereby a color can be represented by three consecutive hexadecimal numbers, each of which represents a color in the RGB color space. RGB is an abbreviation for the proportions of the primary colors red, green, and blue in the mixed color.The usual color definition is in hexadecimal six-digit form, i.e. as a sequence of three two-digit hexadecimal numbers.

[0052] According to another embodiment, the possible perspective of an object can be determined.

[0053] According to a further embodiment of the method, the type and position of one or more light sources of an image can be determined via the prompt.

[0054] According to a further embodiment of the method, the language style of the text - binding, aggressive, clear, binding and / or romanticizing etc. - can be determined via the prompt.

[0055] According to a further embodiment of the method, the vocabulary of a text can be determined via the prompt.

[0056] According to a further embodiment of the method, the background of an image and / or a representation, for example with regard to the contours, the possible color combinations, or a motto, such as nature, technology, emotion, everyday office life, police, administration or other, can be determined via the prompt.

[0057] According to the inventor's findings, a company logo with recognition value is generated by specifying, using one or more prompts with certain stylistic elements, for example that manageable devices such as smartphones, tablets, laptops, headphones, microphones, cameras are always shown against a similarly styled background with the same color combinations in the same perspective and under the same lighting.

[0058] The same applies to non-handleable products such as power distributors, circuits, vehicles, ships, trains, engine housings, cables, distributors, components of decentralized energy systems, medical technology, components of automation technology. Here too, uniform specifications in the form of digital boundaries that implement the design guidelines and with regard to the perspective representation, the placement of the light source, the background color palette can create a consistency of the representations, which is perceived as a company identification with recognition value among the relevant public.

[0059] The digital recording of the "circumstances" by the system takes place - for example, automatically -

[0060] - by recognizing the identity of the user - e.g. IP address, location detection, recognizing the country setting of the operating system -,

[0061] - by recognizing and reading the user's situation and / or emotions, - for example by means of optical and / or acoustic detection, as well as text, gesture, facial expression, sound and / or speech detection -,

[0062] - a connection with the loT, e.g. what the latest news is,

[0063] - the recording of GPS data, e.g. where in the world the input or output device is located, or the user is currently making an input,

[0064] - sensory and / or digital recording of time, temperature, humidity, season, GPS data, other geographical coordinates,

[0065] - the way the processors are connected to the power grid and / or the Internet,

[0066] - Use and / or type of neural network of the LLM and / or the AI ​​and / or quite simply through a predefined and defined input by the user.

[0067] After recording the circumstances, the process and / or system can, for example, automatically adjust the parameters of the avatar so that the avatar appears tailored to the user and a high level of user acceptance is achieved everywhere, which fundamentally facilitates human-machine interaction and therefore makes it more efficient.

[0068] The method and / or the system are therefore suitable for an automatic content modulation and / or cultural adaptation of a chatbot, which is carried out automatically on the basis of a position determination that is also carried out automatically using corresponding sensory input.

[0069] The sensory input can be provided by GPS data collection of the IP address, by other GPS positioning, by all types of measurement such as temperature, pressure, time, light irradiation, humidity, and / or other measurement as input.

[0070] Alternatively or additionally, the system can be equipped with acoustic and / or optical recording devices that capture the user's emotional state, for example, whether the user is already frowning and thinking when given a "trivial" request by the system, or whether the trivial request is also trivial for the user. This provides the system with information about the user's abilities or "skill level," to which the process and the system can automatically adapt, for example, by selecting appropriate design guidelines.

[0071] The method and the system process all or some of the data collected by sensors and / or measurements and assign one or more design guidelines, style guides, etc. to the result. When the result is output after appropriate calculations by the AI, the system adapts to the user receiving the system's response via the design / manner of speech of a responding bot, such as the avatar and / or with regard to the display design, the action of the machine automatically triggered by the human-machine interaction and / or the acoustics selected by the method.

[0072] The system automatically recognises, for example, the season, the cultural region, the industry, and if necessary even the company whose employee is currently making the entry, and automatically uses the appropriate digital design guidelines in the interface.

[0073] The digital design guidelines, which are automatically implemented in the results and the prompt according to the invention, can be formulated and composed based on the knowledge gained by capturing the circumstances, by accessing a library or "library" in which recurring design guidelines are stored and / or by accessing locally stored design guidelines, as well as by any combination of these sources.

[0074] For the automated implementation of design guidelines, the system has, for example, sensors, access to the IoT and / or the acquisition of GPS data.

[0075] For example, a sales negotiation is to be conducted with a partner from the Arab region. The system recognizes the partner from the Arab region and selects appropriate design guidelines.

[0076] For example, the input by the European user Hans D. is:

[0077] "Evaluation, cost-benefit analysis of flight plans for the region xy" A bot or avatar according to one embodiment of the invention transmits the following text to the Arabic customer Ali C . :

[0078] "Because the competence, prudence and analytical foresight of the admired and respected partner Ali C. in Dubai are so highly valued here in Munich, the friend and admirer Hans D. would like to ask Ali C. whether he could possibly devote his valuable time to the question of whether the specified flight routes and flight times in the xy region were selected in a suitable, sensible and economically profitable manner or whether changes would be advisable and would save money, time and / or fuel?"

[0079] In this way, or in a similar way, the Arab partner, Ali C., would feel engaged by the selected bot or avatar and would devote himself to the task and / or assignment with great motivation. In the other direction, for example, a bot, according to one embodiment of the invention, would translate in the same way, so that Hans D., a resident of Munich, would receive the information concisely and succinctly, which might have been provided in Dubai with more text input than was output in Munich.

[0080] For example, libraries are accessible in the IoT that contain a number of common digital design guidelines for download and are accessible online from the system.

[0081] In the following the method is explained in more detail with the aid of a figure which shows a flow diagram of an exemplary embodiment of the method:

[0082] The figure shows four essential steps for carrying out the process, these are: Sl: Digital capture of the design guidelines in word, image, perspective, lighting, colour and / or sound,

[0083] S2 : Use of these results to define digital boundaries of the possible selection range of one or more prompts and / or to generate digital boundaries for the generative AI to compare with the results calculated by the AI,

[0084] S3 : Input of one or more tasks by a user to the generative AI and processing by the generative AI ,

[0085] S4: Output of the results that lie within the digital boundaries, whereby the number of results is limited by adhering to the digital boundaries either during input via the prompt and / or during output automatically by comparing the results generated by the generative AI with the specified digital boundaries within the generative AI.

[0086] The present invention makes it possible for the first time to use generative AI to generate digital communication media for a specific institute or company with predefined design guidelines. In particular, the present invention also makes it possible to automatically convert design guidelines corresponding to regionally different cultural circles into culturally colored AI results using prompts. This means that a single input to a series of prompts for generating a digital product representation, for example, leads to regionally different results worldwide, reflecting, in particular, the respective national customs and prevailing culture depending on the prompt.

Claims

Patent claims 1 . Method for the automated implementation of design guidelines in the output of a result generated with the help of generative artificial intelligence "AI", comprising the following process steps: - Digital recording of the circumstances of the input and output device, - Selection of appropriate design guidelines and - automated implementation of the appropriate design guidelines into the process from user input through AI calculation to machine output, Implementation of the design guidelines by defining the possible selection palette of one or more prompts and / or by generating digital boundaries for the generative AI to compare with the contents and / or results calculated by the AI, storing the generated digital boundaries within the prompt and / or within the processor(s) on which the AI ​​is running, - Input of one or more tasks by a user to the generative AI - Calculating the possible outcomes that lie within these digital boundaries using generative AI - Output of the results taking into account the selected design guidelines, whereby - the automatic selection of the results by means of the digital boundaries and / or design guidelines takes place either during input via the prompt, during calculation by the AI ​​and / or during output by automatically comparing the results generated by the generative AI with the specified digital boundaries and / or with the specified digitally available design guidelines within the AI.

2. Method according to claim 1, wherein the digital recording of the circumstances takes place via sensors, GPS data and / or connection to the IoT.

3. Method according to claim 1 or 2, wherein the digital recording of the circumstances automatically results in the generation and / or retrieval of design guidelines in word, image and / or sound in a memory module of the generative AI. 4 . Method according to one of the preceding claims, wherein the results of the generation of design guidelines in word, image and / or sound are stored in a memory module.

5. Method according to one of the preceding claims, wherein the design guidelines specify a color palette. 6 . Method according to one of the preceding claims, wherein the design guidelines specify a perspective.

7. Method according to one of the preceding claims, wherein the design guidelines specify style elements typical of the company. 8 . Method according to one of the preceding claims, wherein the design guidelines specify country-specific and / or industry-specific stylistic elements.

9. Method according to one of the preceding claims, wherein the design guidelines specify a language and / or a language style.

10. System for carrying out a method according to one of claims 1 to 9, wherein the system has an interface which has means for recording the circumstances and for the automated selection and implementation of design guidelines, as well as prompts for requesting and / or input and forwarding to the generative AI, wherein the interface between the user and the processor on which the AI ​​is running is provided in such a way that this interface is coupled to the prompt, so that the prompt is set up and configured by means of the interface in such a way that it only forwards inputs that are within predetermined digital boundaries to the generative AI, whereby variants and possibilities from the input that are not within the digital boundaries are identified and removed and / or deleted by the prompt.

11. The system of claim 10, wherein the digital boundaries are based on generated and / or stored design guidelines.

12. System according to one of claims 10 or 11, which has sensors, access to the IoT and / or for acquiring GPS data. 13 . System according to one of claims 10 to 12 , which has access to a library comprising a number of design guidelines.

14. Computer program product with program code means which cause an electronic computing device to carry out a method according to one of the preceding claims 1 to 9 when the program code means are processed by the electronic computing device. 15 . Computer-readable storage medium with a Computer program product according to claim 1 .

Images