Colourful robot images generated with Midjourney | AI in Design | PESCHKE
AI in design: From hype to legal concerns and new possibilities

For a while now, countless articles, opinions, utopian imaginations and dark predictions of various horror scenarios have been running rampant on the internet, all regarding one topic: the use of artificial intelligence (AI) in design, as well as related existential fears of designers and legal concerns regarding copyright, ownership and what art or design even are.

We studied the subject intensively and take a strong stance: AI can serve as a great inspiration, as a basis to start from and it is one of many tools we as designers use. AI, however, will not replace good designers.

With this article, we want to show the potentials of AI in creative applications, create awareness for legal matters and position ourselves against the fearmongering that is currently taking place.

April 18, 2023

Digital design
Close
Digital design

Less in the sense of artificial intelligence and more in the sense of augmented intelligence », AI can simplify a range of design processes. Instead of reinventing the wheel with each design project or researching which page navigation system meets the latest standard, we can use AI to complete such monotonous tasks. AI can, for example, do the research for you and even list useful guidelines.

This presents designers with the opportunity to focus on the truly important aspects of the design and to think innovatively.

Use cases

Text creation
Close
Text creation

Who hasn’t had to deal with it: Numerous e-mails waiting to be written and answered, but neither enough time to do so nor the creativity to know how best to write them. After all, there are already enough other tasks waiting one one’s desk. Or topics for social media and blogs that you first have to research before you can start writing about them.

There are plenty of examples for required text content, but by now, there are thankfully equally as many text creation programmes for various content that take off some of your work. Which, as mentioned, leaves more time for you to dedicate to the truly important stuff (those who require something less strenuous to do between all their important tasks might want to try Townscaper » – just a little tip).

The probably best-known text creation software is ChatGPT ». The programme delivers surprisingly well-formulated texts that, for the most part, feel as if written by humans. But in reality, the AI does not do much more than put one likely word after the next based on calculated probabilities (with some variation to seem more human). Well, it is a little more complex than that; for those who are really interested, we recommend this article on how ChatGPT works ».

Use cases

There are countless ways you can use the software. They reach from simple text creation to preparing copy text for prototypes to market research that forms the basis for product development. Here are two specific use cases:

Video game development
Close
Video game development

AI has long been used for the creation of video games in different ways. But the new use of AI for the creation of elements such as background sceneries, voices, music or narratives raises concerns regarding copyright. This is because AI, similarly as in other fields of art, cannot make great new games appear out of thin air without referencing human creations as it lacks creativity and does not think the way humans do. Apart from these concerns, however, AI increasingly supports areas such as error detection » of videos, among other monotonous tasks, meaning that game artists can focus on the complex parts of game development. So as long as AI is used to support the actual video game artists, it brings great advantages and contributes to the improvement of games and work flows.

Use case

Product design
Close
Product design

Product designers can definitely not be replaced by AI yet. Designs that are created based on existing designs may serve as inspiration, but they do not come with a technical feasibility check, nor an optimisation of ergonomics, nor do they pay attention to emerging topics such as circular design. In addition to their use as a source of inspiration, AI has another potential application in product design. At the moment, it is still a dream of the future, but AI could greatly facilitate matters as a source of knowledge for complex topics in product design, such as the building envelope ». Knowing and complying with all standards and regulations, always meeting the state of the art, and all this from a financial, ecological and social point of view, is virtually impossible and yet expected. Researching materials and material properties alone is a huge project in itself. Those who fail to meet requirements will, in the worst case, have to deal with the issue of product liability. A well-functioning AI can provide support in this area by covering research, listing requirements and thus ensuring compliance with relevant regulations.

Use cases

Other creative areas & general benefits
Close
Other creative areas & general benefits

Of course, it does not stop with the above-mentioned areas: Music, programming, architecture and all other areas will feel the impact of AI ».

What all areas will benefit from is the possibility to generate ‘checklists’ with AI so that you can carry out creative tasks freely without having to worry about missing standard elements or similar. Designers and artists can also use AI to compare their designs with AI-generated suggestions – just as you would do when comparing with colleagues or the competition.

A special potential of AI is the fact that –provided it gains access to all the writings and other works ever created by humans– it can identify new connections from the billions of pieces of data that humans were previously unable to make. After all, it is impossible for a single person or even groups of people to know and combine all the data in the world on a specific topic. It can therefore be assumed that correlations discovered through AI will enable great leaps in a variety of areas.

Colourful robot images generated with Midjourney | AI in Design | PESCHKE
AI-generated robots in various styles depending on the entered prompt
AI-generated robots in various styles depending on the entered prompt
AI Images generated with Lexica | AI in Design | PESCHKE
Extra arms, fingers and paws: AI is not completely faultless yet
Extra arms, fingers and paws: AI is not completely faultless yet
Copyright & property
Close
Copyright & property

Probably the most hotly debated topic in relation to AI, especially in the creative field, is the issue of copyright infringement and copyright law. After all, AI outputs are based on examples of real artists that AI has been trained on – without having asked the creators for their permission or offering them a share of the profits. But should creators automatically earn money as soon as their art is used for learning, or only/also when related AI art is sold? According to experts, a possible solution would be to compensate artists on a contractual basis for the use of their images by AI generators ».

And who does AI-generated art belong to? The programmers? Those who generate them through prompts? Certainly not the application itself, as non-humans cannot have ownership. Because of all these unresolved questions, there has already been a class action lawsuit by several artists in the US » against Stability AI, Midjourney and the platform Deviant Art, as these programmes used their art for machine learning without permission and without a share in profits.

The current loophole is that AI-generated creations are available free of charge to anyone and that the creators of the AI software thus do not make any money from them either. However, the situation is different when someone takes AI-generated materials and uses them for commercial purposes. It suddenly becomes very difficult to understand who is earning money from the work of other artists and where boundaries need to be drawn.

In our exploration of these questions, we received answers to the most burning legal questions from design law expert Dr Markus Grötschl. You can find these answers in the video interview below this section.

A temporary solution?

There will certainly be new laws in the course of the next couple of years, but until there is a legal framework, artists can at least protect their art with Glaze » to stop AIs from learning from their digital or digitised work. The programme covers images with a ‘filter’ that is invisible to the human eye », but makes the content of the images unrecognisable and thus ‘unlearnable’ to AIs. Another interesting aspect is the connection between images, AI and SEO optimisation: From an SEO point of view, it makes sense to provide images with descriptive alt tags so that they show up in (Google) search queries. However, it is precisely these alt tags that make it very easy for AIs to track down images for certain keywords and include them in their ‘repertoire’.

There is already a lot to think about in terms of copyright, but there is another open question: What about copyright on work created with artificial intelligence? According to US federal law, copyright only applies to work created by humans or with their input ». This means that machines cannot be contractors under copyright law, as they would either have to enter into an employment relationship or sign a legally binding contract, which of course a machine cannot do.

Laws can change, but since the current copyright loophole of AI services is to provide AI-generated creations for free, AI-savvy artists would do well to consider the lack of copyright claim if they want to use AI for their own art.

However, some experts are optimistic and are sure that artists will soon be working with AI instead of against it ».

'Fake news‘ and untruths
Close
'Fake news‘ and untruths

“Deepfakes” caused a stir a while ago, as the ability to depict (made up) people in (made up) environments or contexts in photos or even in videos can be used to spread false claims as they can often not be identified as “fake” ». Likewise, this can happen, albeit unintentionally, with the output of text creation software as it is not good at judging the truthfulness of a text it created. Recently, there were scandals surrounding ChatGPT, as the programme had only been trained with data up to 2021 » and was therefore unable to accurately answer queries on current topics. In addition, the large amounts of data with which AIs are fed already contain misinformation from humans, which is passed on.

The general trust in the internet is relatively high among many people (e.g., Google search results on the first page are automatically trusted the most ») and “fake news” are often not recognised by certain groups. People who want to ease their workload and therefore use ChatGPT to quickly generate texts on a certain topic will not always question facts, but will gratefully accept them. If programmes like ChatGPT are to be used on a large scale, these limitations must therefore be taken into account.

Biased? The influence of programmers
Close
Biased? The influence of programmers

AI consists of code made by humans. AI-derived output is also heavily influenced by what the AI has been trained on and how algorithms process that content – which even has implications for court decisions » as artificial intelligence makes decisions with the biases of its developers and of those who created the materials it learned on. For example, AI contributes to the perpetuation of discrimination.

For a long time now, content and advertisements on social media channels have been selectively fed to us by algorithms and alter our perceptions. In the same way, the views of programmers will influence AI-generated output in the field of art, thus indirectly influencing us with their own perceptions of the world, prejudices and political judgments.

When analysing AI art in this regard, you can see clear tendencies in different programmes as to which default styles they use when short prompts are entered. This means if you do not specifically try to achieve a different end result, you will be strongly influenced by the programmes’ defaults.

The missing human component
Close
The missing human component

AI can do many things, but what it does not have are human creativity and compassion. The lack of a human contact and consultation is therefore a major disadvantage of AI in the creative industry. This may not matter when buying AI-generated posters online for home decorating, but it does matter a lot when implementing marketing campaigns, product designs or other projects on a larger scale. Many people are simply frustrated when they cannot turn to someone who understands them on a human level when they face problems or want to request changes.

From our point of view, designs developed by humans have the essential advantage that they take into account all those things that AI cannot. These include emotions, to a large extent, but also the courage to try something new as well as intentional provocation. In any case, the professional consultation of our clients and the development of innovative designs is clearly better suited to us designers.

AI applications in other sectors
Close
AI applications in other sectors

With AI, vast amounts of data can be collected and analysed, which can bring great benefits and efficiency in terms of smart connected products and the IoT (Internet of Things) ». At the same time, however, there is a danger that corporations will misuse this data and interfere with the privacy of their customers in order to monetise the insights gained ». This is where data protection, human rights, privacy and security become problematic issues. The development of AI must therefore respect human values, including privacy, in order to be truly intelligent and meaningful.

  • Technological trigger: It all starts with a potential technological breakthrough that gets a lot of publicity through (premature) proof-of-concept reports and media coverage. At this stage, however, there are –in most cases– no viable products or proof for commercial use yet.
  • Peak of inflated expectations: Due to early publicity, we hear of countless success stories, but also of just as many failures and problems. Some companies are immediately fired up and take action, but many resist or think it is nonsense.
  • Trough of disillusionment: Now that it is common knowledge that several experiments and implementations were unsuccessful, general interest wanes. The manufacturers of the technology fail. The remaining providers who still want to stand a chance of investment must improve their products to the satisfaction of the early adopters.
  • Slope of enlightenment: The benefits of the technology are increasingly well understood and more and more use cases emerge for companies. Technology providers release second and third generation products and more companies invest in pilot projects. More conservative companies still act reluctantly.
  • Plateau of productivity: The product evolves from a niche product to a mainstream product. Providers can evaluate the profitability of the product with clearly defined criteria. The broad market applicability and benefits of the technology are profitable.
Gartner Hype Cycle | PESCHKE Design Studio

Let's turn your vision into reality!

    Make an appointment right away?

    You’d prefer to contact us directly? Of course! Here’s how to reach us:

    You’d prefer to contact us directly? Of course! Here’s how to reach us:

    Mail: contact@peschke.at

    Tel: +43 1 47 07 922

    Make an appointment right away?

    Sure! Directly book a non-binding initial consultation to share your vision and ideas with us – free of charge, of course. We look forward to meeting you!

    Book appointment Request appointment
    Many thanks for your message!
    Your message has successfully arrived in our inbox. We are thrilled about your interest and will contact you shortly to discuss your project ideas.
    Back to top Arrow