Fundamental questions about artificial intelligence
- Hopes and fears of artificial intelligence
- Copyright
- Quality and originality of AI-generated images
- Ambiguity between creator and consumer
- Loss of the pleasure of creating
- Transforming creative professions with generative AI
- Addiction to artificial intelligence
- Cognitive loss in child learners
- Closed-loop creation
- Responsibility for AI-generated content
- 50 arguments against using AI according to ChatGPT
- Are we using generative AI at AOKIstudio?
Hopes and fears of artificial intelligence
Artificial intelligence raises both hopes and fears, and opinions on the long-term consequences of its use are extremely divided.
Will artificial intelligence save humanity? Will it solve problems, such as climate change, that humans are no longer able to resolve?
Will AI play a role in protecting the environment, optimising the use of resources and developing cleaner, more sustainable technologies?
Will future super-intelligences that surpass human intellectual capacities help us to understand the origin of the universe and propose a better world?
Are we at the dawn of an era of positive collaboration between man and machine?
Or, on the contrary, does artificial intelligence signal the end of human intelligence? Does it foreshadow a world where learning and knowledge are no longer necessary?
Will future generations of artificial intelligence have feelings and desires? To guarantee the survival of the planet, and therefore their own survival compromised by the uncontrollable actions of humans, will the machines of the future choose a future without humans?
These existential questions, which are very important for the future of humanity, sometimes overshadow more immediate concerns such as copyright, the distinction between creator and consumer, the loss of the pleasure of creating, addiction, the cognitive effects on learning children, the future of a closed-circuit creative process and the responsibility of AI-generated content.
Copyright
The use of AI by the general public and professionals in creative fields such as writing, design, music or the visual arts raises many questions. Some countries such as Italy, Russia and China, for example, have banned ChatGPT for privacy issues and spreading false information.
When it comes to copyright, generative artificial intelligences such as ChatGPT and Midjourney create images and text by training their AI models from copyrighted data and works, but governments are helpless in the face of these ubiquitous copyright infringements in all creative fields.
In the fields of design, painting and illustration, software such as MidJourney, Dall-E or Stable Diffusion make it possible to tag artists' names and are particularly destructive. Contemporary artists see their work plagiarised and re-used without authorisation by the general public and professionals alike. Far beyond potential competition from AI and the loss of their work as artists to AI, seeing one's artistic work plagiarised, modified, distorted and used without authorisation causes a feeling of discouragement and great sorrow.
For example, the character of Totoro created by Hayao Miyazaki has been tagged so often in Midjourney prompts that it is no longer possible for the general public to tell the difference among the images available on the Internet between the original character and the copies.
It is astonishing to note that many professionals in fields as diverse as design, fashion and advertising tag the names of artists on AI image generators without considering the damage this causes.
Artists, authors, rights holders and beneficiaries of copyright are seeking to change this situation.
Among the measures proposed are transparency of AI models and a requirement to obtain authorisation from authors before including their works in AI platform databases. Not attributing intellectual property to a work created by a machine, requiring all works created by AIs to carry a statement such as Created by such AI platform are also priority measures.
On the contrary, AI companies and their users advocate cultural democratisation, access to artistic creation for as many people as possible, including those with no training in a creative field, and invoke the legal concept of fair use, an exception clause to copyright which in many countries allows both the interests of authors and the public interest to be taken into account by authorising the use and dissemination of copyrighted works for the purposes of criticism, comment, journalistic information, teaching and research without needing the permission of the rights holder.
However, the main criteria for determining whether it is a fair use are the purpose and nature of the AI use, including whether it is commercial or educational and not for profit, and the impact of the AI use on the potential market for or value of the protected work. The use of AI, particularly by communications agencies and brands for marketing and sales purposes, therefore already seems far removed from the framework of fair use.
It is therefore imperative that AI technology and machine learning models are developed responsibly, with respect for copyright and intellectual property protection.
Copyright legislation and policies must evolve, take into account the impact of AI on cultural creation and consumption and guarantee a balance between access to artistic creation for the greatest number and the protection of authors' interests.
Quality and originality of AI-generated images
Does AI, without understanding the content it generates, simply reproduce existing patterns and structures in its training data, generating content that lacks originality or quality? Is AI, through the ability of its algorithms to randomly merge and reorganise information, capable of generating unique and innovative content that equals and sometimes surpasses the human imagination?
Opinions are divided as to the originality of the texts and images generated by artificial intelligence, but they are not contradictory.
Indeed, artificial intelligence systems, like language models or image models, have no consciousness, sense of understanding or real intention, they simply follow the instructions encoded in their algorithm, which are based on the patterns they have identified in their training data but, by combining information, mixing and matching elements of their training data in a different way to humans, they generate surprising results of great originality.
AIs like clouds have no consciousness or representation of the world
AI generates content randomly on the basis of its algorithms without deliberate intent, will or particular understanding and it is humans who interpret and give meaning to that content. In this context, AI generates content that humans can consider original and inventive, images, texts, sound worlds but also concepts that would probably never have existed without AI.
AI does not decide or know whether what it generates is original or inventive. When an AI model generates a text, a piece of music or an image, it is the reader, the listener or the viewer who attributes value and meaning to it; it is they who can be touched by the beauty or originality, feel an emotion that goes beyond the algorithm.
We could take the shapes of clouds in the sky as an example. Clouds form as a result of natural processes in the atmosphere and do not create shapes intentionally; it is humans who attribute a resemblance and meaning to the shapes that certain clouds take on.
"Thanks to art, instead of seeing a single world, ours, we see it multiply, and as many original artists as there are, so many worlds at our disposal, each more different from the next [...]."
Le Temps retrouvé, Marcel Proust, 1927.
Cloud objects
Desing : Olivier Defaye, AOKIstudioAmbiguity between creator and consumer
With the emergence of AI algorithms, particularly in the field of artistic creation, we're living through an incredible moment, certainly a revolution like the mastery of fire or the invention of printing.
For many, a magical moment where the pleasure of generating concepts and images becomes accessible to all but the incredible originality and quality of the texts and visuals generated by AI does however create an ambiguity between the status of creator and that of consumer.
Users' vocabulary tends to signify an appropriation as creator, as authors of the concepts, texts and images generated with AI.
For example, look at what I've created, it's my work, I did many tests, many combinations of keywords before creating my image....
With this tool, we draw our illustrations three times faster!
I scored 20/20 for the assignment I did with ChatGPT!
The viewers' vocabulary is also revealing: I love your work, you're my favourite AI artist....
While we are aware that our input is limited to a few words, the tools of prompt art elicit a strong sense of creativity.
"If I hadn't written those words or chosen that combination of words (which I secretly guard as an admirable skill), that concept or image wouldn't exist."
That's true! But the aesthetic results and concepts generated by AI from a few keywords go far beyond what we could have imagined.
Proposing keywords and visual references without having an idea of the result a priori, reformulating these keywords according to an unsatisfactory result, corresponds, in our traditional world, to what a client (an advertising agency, a brand, an NGO, an art gallery...) might ask a creative director, an art director or an artist to obtain a concept and an image from their brief.
In the case of prompt art, the person generating an image reformulates their request with new keywords to arrive at a result that suits them as systematically as a client in the working world. So would our status be closer to that of a client than that of a creator or author?
The status of client is perhaps not the most relevant when we generate images with AI. At least not the client of an advertising agency or an art gallery, but rather the customer who buys a box of nuggets thinking he has chosen his meal and without knowing what it contains or the process of its production, in other words a consumer.
When we ask AI to generate images, aren't we simply consumers?
A box of nuggets please! with Hawaiian sauce and greens.
Check out ChatGPT4's response to the fact that AI users are consumers, not creators.
Loss of the pleasure of creating
Do we play computer-assisted chess with the aim of winning, or do we play without assistance for the pleasure of the game?
Are we going to ask our robot to play tennis for us because it's better than us?
Interesting questions to ask users of AI tools in creative fields!
It has to be said that the creative results achieved by AI are impressive and have surpassed or will most certainly surpass those achieved by humans. But is the creative process only about the result?
What happened to fun in the creative process?
Can generating an image or text from a few words be a source of pleasure?
Has the pleasure of creating disappeared in favour of the result alone?
Why are so many designers converting to artificial intelligence? Some artists and designers are already saying they can't live without it! In admiring the results of ChatGPT or Midjourney, have we lost our taste for creation, perseverance in the face of challenges, risk-taking and the immense satisfaction that comes from discovering something that suddenly emerges, as if by magic, from our imagination? Could it be that the fascination with the infinite capabilities of the human brain is being replaced by a fascination with machines?
Have we lost our taste for challenges?
In many fields, it is wonderful that AI is replacing laborious and unrewarding tasks previously performed by humans. But in creative fields?
Is the creative act so difficult that we are ready to abandon it as soon as a technological alternative appears?
Advocates of AI in creative fields see it as a tool to stimulate the imagination, help come up with new ideas, creative solutions, help overcome creative blocks, explore new avenues and push the boundaries of creativity.
YES!
But why?
Where's the fun in that?
The priority given to the result, its quality or its speed of execution, to the detriment of the pleasure of the creative process, naturally leads to a loss of the desire to create. Why create when AI can do it for you?
What happened to the satisfaction of having explored with difficulty and then found an original idea or concept, a design or a personal artistic universe?
As well as making us lose the pleasure and desire to create, can AI also shake our self-confidence?
I don't feel confident in this game of chess because my opponent is so much stronger than me.
But never mind, let's play!
Check out the response from ChatGPT4 about AI in creative fields and the loss of the pleasure of creating.
Cover of CG word magazine, Japan
Desing : Oliver Defaye, AOKIstudioTransforming creative professions with generative AI
Artificial intelligence, like all new technologies, is leading to the disappearance and transformation of certain professions and skills, as well as the emergence of new professions.
In scientific fields such as medicine and mathematics, Isabelle Ryl, a researcher at Inria, points out on Radio France that "AI replaces humans in automatic tasks so that they can concentrate on their real added value."
In artistic fields such as visual arts, product design, architecture, literature and music, generative AI is also replacing laborious human tasks and automating certain repetitive and time-consuming parts of the creative process. In the field of image processing and photo retouching, for example, AI can take on tasks such as correcting exposure, adjusting tints, reducing noise and creating masks. AI can also identify and separate the subject of a photo from its background, making it easy to replace the background.
3D modelling tools already incorporate AI to facilitate the creation of shapes, the deformation of objects and the creation of textures.
AI can generate 3D models of existing objects from 2D images or optimise the number of polygons in 3D models.
AIs such as LDM3D (Latent Diffusion Model for 3D), developed by Blocklade Labs and Intel Labs, can generate, 360-degree view and export to specialist 3D software realistic 3D environments from text prompts (LDM3D demo...) have the potential to revolutionise content creation, metaverse applications and digital experiences in industries such as entertainment, design and architecture.
In terms of 3D animation, an AI trained to understand how a human character moves naturally will also be able to automate parts of the animation process, such as realistic character movements.
But will AI allow artists and scientists alike to focus on their real added value?
Will AI be used as a tool to help artists express their true creativity?
Will it replace fulfilling tasks in creative professions where originality and personal creativity were highly valued?
Are we going to outsource creative work to generative artificial intelligences? Is AI an assistant or a subcontractor?
We leave it seems, generative image models like Midjourney, DALL-E or Stable Diffusion, gradually take on the most rewarding and fulfilling artistic roles, the most stimulating and challenging tasks.
A growing number of creative professionals, such as artists, art directors, architects, designers and graphic artists, are delegating the most creative aspects of the creative process to AI, such as developing concepts, ideas and designs because the technical and artistic quality of the results generated exceeds their creative abilities.
These professionals abandon their imagination and hide behind the art of the prompt, presented as a skill that is difficult to master and the secret of a new talent.
Some people use the term "assistant" when referring to AI tools, but wouldn't the term "subcontractor" be more appropriate?
IA, assistance or subcontracting?
Anyone who has tested image-generating AI has found that, although a few keywords are enough to produce impressive results, these results do not correspond to the original intention formulated at the time of the query and are often very different from what might have been anticipated.
But it doesn't matter! It's different from what I wanted, but of much better quality than what I could have come up with myself: I'm happy with it!
Shouldn't we simply accept the fact that we are giving up our creative power, our imagination and the pleasure of discovering a work that has magically emerged from our personal experience, our education, our culture and our feelings?
Check out the response from ChatGPT4 about the transformation of creative professions with AI.
Addiction to artificial intelligence
An addiction naturally arises when we regularly consume or use the same product or tool.
Many designers, architects, stylists and graphic artists are already talking about their addiction to artificial intelligence.
I already can't do without AI!
Have AI tools become indispensable to our working lives?
In fields such as design, architecture, fashion or graphics, AI facilitates and accelerates the creative process, but a thoughtful use of AI would be to consider it as a tool to assist creation, a tool to stimulate the imagination for example.
However, the quality of the texts and images generated by AI does exert a certain fascination and invites us to use the results generated as they are, as a turnkey solution.
In creative fields, AI generates texts and images that are more often than not used as they are, rather than as food for thought or sources of inspiration. If the result generated does not meet expectations, the user does not try to develop it based on personal intuition or imagination, but simply modifies the keywords in the prompt.
The user is no longer the main actor, letting the AI take control of the creative process and becoming dependent.
Does dependency have an impact on personal well-being? Open answer.
What if Paris was alive?
Desing : Olivier Defaye, AOKIstudioCognitive loss in child learners
Does the use of AI affect our natural abilities to conceptualise and imagine?
Will we lose our power to imagine, conceptualise and understand?
The appearance of new tools, such as the combine harvester in the mid-19th century, for example, is accompanied by the disappearance of the manual labour, skills and trades that these tools replace.
Few people knew how to make a fire as our ancestors did, using only twigs.
It is legitimate to ask whether the adoption of tools that replace part of our memory, cognitive and imaginative abilities could lead to the disappearance of certain trades, but also, and above all, to the disappearance of natural abilities, skills and know-how.
Do calculators weaken our mental arithmetic skills? An open question!
The question of the impact of the Internet on our memory is still topical.
Does the constant availability of information online affect our memory?
Does the externalisation of our memory on the Internet and the fact that we no longer have to memorise information ourselves affect our ability to retain and remember? Does it affect other cognitive abilities?
Or, on the contrary, does the Internet free up our memory and cognitive capacity for creative or critical thinking?
What will be the consequences of children using AI tools in their learning?
Some believe that AI will make it possible to offer a fun and personalised environment conducive to learning, adapt to individual rhythms and facilitate access to education for many children.
Others are thinking about dependency, the excessive use of AI tools to the detriment of activities necessary for children's development, such as outdoor play, social interaction, and non-digital creative activities.
Preventing our brains from an opportunity
In the same way that we entrust our memory storage to the Internet, when we delegate a task to a machine, are we depriving our brain of the opportunity to practice and strengthen the skill needed for that task? If a child generates texts, calculations, drawings or music with AI rather than by himself, won't this have a negative impact on the development of his writing, drawing, calculation or music skills?
If young child learners use AI before they have developed their ability to conceptualise and formulate ideas, will they ever be able to think for themselves and live without AI?
The Internet may enable us to develop our creativity and critical thinking by freeing our brains from the need to memorise, but will the use of artificial intelligence by child learners enable them to develop their natural faculties of conceptualisation and invention?
Is this already the end of human inventiveness and intelligence?
PS: We are all big children... learners.
Closed loop creation
Machine learning models, such as image generation AI or language models, were initially trained from human productions.
With the increasing use of AI in creative fields, the number of images and texts generated by AI will quickly surpass that of human creations.
Given the impossibility of distinguishing between data generated by AI and human creations, artificial intelligence tools in creative fields will no longer train solely on data created by humans, but also on data generated by themselves. It is possible to imagine that the future databases that will train AIs will be mostly made up of AI-generated data, with the amount of human-generated data becoming extremely small in comparison.
Will an AI image generation tool, re-trained from the images it has itself generated, enable new inventions to be explored and introduced into creative fields? Will it lead to a closed-loop creative system and a loss of diversity and innovation?
Check out the response from ChatGPT4 about the closed loop of creation with AI.
Who is responsible for content generated with AI?
Responsibility for content generated with AI is a complex subject that is the subject of much legal and ethical debate.
Who is liable when content generated by AI tools does not respect copyright? Who is responsible for defamatory messages, fake news, violent content or content that breaks the law?
Are those responsible the AI developers who have not put in place measures to prevent copyright infringement or such results?
Are the users who, through their choice of prompts, are responsible for the content generated?
Are authorities and governments, unable to put laws and regulations in place in a hurry, responsible for the content generated?
The legal and regulatory framework determining responsibility for AI-generated content is still in its infancy and will no doubt be constantly evolving.
Check out the response from ChatGPT4 about responsibility for content generated with AI.
50 arguments against the use of AI according to ChatGPT
The impact of AI on art and human creativity led us to ask ChatGPT4 to generate 50 arguments against the use of AI in creative fields.
Check out the 50 ChatGPT arguments against the use of AI in creative fields.
Are we using generative AI at AOKIstudio?
We're always on the lookout for new tools and technologies, so we test text and image generative AIs with great curiosity.
We were impressed by the quality of the results, the ease of use and the speed of generation, but there were several points that put us off adopting AI.
Among the main reasons, plagiarism of existing artists and losing the fun of the creative process.
These 2 points alone could be the reason for not using generative AI, but another observation supported our decision.
Loss of control over the creative process
AI tools are presented as creative aids, tools designed to generate ideas, new aesthetic directions, enrich and facilitate the artistic process.
We have carried out numerous tests with different generative AIs and have simulated their use in professional contexts such as proposing an advertising concept or a graphic universe as part of the production of animated films. As well as raising questions about the authenticity and originality of the results generated by the AIs, we quickly got the feeling that the AI was insidiously taking control of certain creative decisions, creating a feeling of creative dispossession.
That's great! But it's not what I wanted at all.
In the context of graphic research, which we would have carried out before the advent of AIs on the basis of references taken from art history, photos or sketches by hand or in 3D, we realised that the results of images generated by AIs were distracting us from our initial intention.
Indeed, the results generated by the AIs are all the more spectacular in that they rarely correspond to what we had in mind when we wrote the prompt.
However, the quality of the results encourages us to choose one of the options proposed by the AI, rather than taking the time to look for a more personal creative direction.
These results, which are so accomplished, seem to be imposed on us and limit our creativity, giving us the impression that we are choosing a direction by default, rather than out of a genuine creative desire.
The results generated by the AI, which we initially thought we would use as research and brainstorming tools, quickly turned into a turnkey solution that we offered to our customers. We felt like we were becoming supervisors rather than creators, becoming our own customers.
We follow the development of generative AI with great curiosity and continue to test these new tools on a daily basis for research and development purposes, but we have made the choice not to fall into the easy way out and dependence on generative AI in a professional context out of respect for our customers and partners.