In a world where speed, affordability, and quality are often at odds with each other, artificial intelligence (AI) seems to be the perfect solution. AI is revolutionising the way we write, code, compose, and draw. But before you start panicking about robots snatching up all the art jobs, it still has a way to go before it can match human artistic ingenuity.
The rapid development of generative AI technology has left many wondering what impact it will have on the future of commercial art. It’s also rekindling concerns about data protection, copyright laws, and business ethics.
The Industrialisation of Art
Some of the discussion around AI echoes the concerns of artists centuries ago during the Industrial Revolution.
The Industrial Revolution of the 1800s was a time of major mechanisation and innovation. Britain transitioned from an agrarian economy to a manufacturing economy. From a mostly rural society to an industrial society. From creating goods by hand to using machines.
Not everyone was happy with industrial modernisation. The Arts and Crafts movement was born out of the belief that machinery and factory production was to blame for the decline of standards in the decorative arts.
The movement comprised many prominent figures of the time, whose influences are still felt today, including art critic John Ruskin, writer Thomas Carlyle, craftsman-designer-poet William Morris, and artist-designer-architect Charles Rennie Mackintosh.
Arts and crafts practitioners questioned the moral, social and aesthetic merit of industry[1].
They took offence to the use of machinery – how it triggered the division of labour and the loss of traditional craft methods. They believed factory-made work to be artificial, soulless, and dishonest.
However, the Industrial Revolution led to increased production, lower prices and higher wages. Beautifully crafted furniture, too expensive for average families to buy, became the inspiration for more affordable mass-produced furniture pieces. In turn, it amplified awareness and appreciation of authentic handmade goods.
The outcry and outcomes of the Industrial Revolution are common in most technological innovations. A push and pull of winners and losers. As old occupations decline, new occupations emerge.
Jump cut to present day and we are amidst this outcry. Tech companies and governments are pushing for rapid change while artists are, once again, challenging them on their moral and social responsibilities.
Machine Learning or Machine Stealing?
Many of the leading AI companies have trained their models using web-scraped datasets. These are mostly uncurated and are known to contain personal information and copyrighted material. Why is this allowed?
Due to the volume of data needed to train an AI, copyright laws grant exemptions that allow noncommercial researchers to make reproductions of copyrighted work for Text and Data Mining (TDM). TDM being a catch-all data processing term that includes Machine Learning and AI. This exemption helps researchers to work fast, staying competitive.
The only stipulation is reproductions are given sufficient acknowledgement…unless it’s not practical.
Furthermore, everything enters a grey area if the data is passed on for commercial purposes. An act referred to as “data laundering.” Numerous AI companies have either funded or set up non-profits for research purposes, to later monetise the models through for-profit companies.
On a side note, scraping data is its own contentious issue. Though technically legal, it has forced social media platforms to put in measures to combat being scraped. Many platforms now limit the content viewable if you’re not logged in. Some explicitly prohibit it in their terms of service. However, that does little to deter scraping companies.
Current Copyrights Protection
For some artists, AI-generated artwork is uncomfortably similar to their own. And, at the rate of which the AI artwork is generated and shared, soon the original artwork will be eclipsed.
Logically, artists look to copyright laws for protection.
For most countries, copyright gives a person exclusive rights to their original tangible work, acting as a deterrent from exploitation. It is given automatically, no need to apply or pay, and lasts the owner’s lifetime, and then some[2][3].
Sufficiently distinctive characters qualify. However, ideas, creative techniques, and style of art do not constitute something ‘tangible’ so are not eligible for copyright protection.
In addition to the TDM exemption, there are also some more copyright caveats, that allow others to use copyrighted work for commentary, criticism, parody, or transformation.
Generative AI opens up many questions around copyright; Is the data input infringement? Are the AI-generated works sufficiently unique or transformative? Does machine-made work qualify for copyright protection?
The ambiguity around outdated copyright laws means that it’s up to the courts to provide answers to these questions in lawsuit, after lawsuit, after lawsuit.
Steps Being Taken
All this flagrant use of data has people scrutinising terms of service on every platform they use; smashing that “opt-out” button whenever they see it.
With a lack of data safeguards, unresolved questions around AI’s ethics and legalities, many creators are voicing their concerns and some have banded together.
Concept Art Association, headed up by Karla Ortiz, is ‘fighting back against the unethical practices happening in the AI text-to-image space.’ The European Guild for Artificial Intelligence Regulation is an initiative by artists from all over Europe ‘united in bringing to the public attention how our data and intellectual properties are being exploited without our consent.’
Creative agencies and associations are advocating members’ rights and are liaising with policy-makers around copyright permissions, and financial remuneration[4].
Tech companies are also taking steps to make the fear of the unknown a little less scary.
There is a promising unified effort between The Coalition for Content Provenance and Authenticity (C2PA) and the Adobe-led Content Authenticity Initiative (CAI) advocating for responsible digital media creation and sharing. They’re focused on open technical standards that can provide context and history for different types of digital media.
The need for provenance and attribution of our digital content has never felt more important.
Team Human
Similar to how the Industrial Revolution motivated craft makers to expand their skills, we are constantly seeking out new tasks. Ones that machines cannot perform yet. We strive to become experts in these areas so that machines cannot easily catch up.
Even if we stop finding human-only tasks, no level of technological advancement can replace our affinity for the human touch. It’s why we have a more visceral response to practical effects than to CGI. Why we admire printmaking over an inkjet print. It’s why observing excellent digital art on a screen will never have the same impact as viewing an exceptional painting in a gallery. It’s also why craft is still big business and traditional craft skills are safeguarded.
Your taste and choices, knowledge and experiences, personality and passions, everything that makes you unique, is the real value. The more honest and brave you are, the less pressure you’ll feel as the industrialisation of art presses on.
References
- “Arts and Crafts movement”, Wikipedia, February 25, 2023, https://en.wikipedia.org/wiki/Arts_and_Crafts_movement
- “How copyright protects your work”, GOV.UK, February 25, 2023, https://www.gov.uk/copyright
- “What is Copyright?”, Congress.gov, February 25, 2023, https://www.copyright.gov/what-is-copyright
- “Artificial Intelligence and images”, The Association of Illustrators, July 7, 2022 https://theaoi.com/2022/07/07/artificial-intelligence-and-images
Further reading
Articles
- OpenAI Blog — DALL·E: Creating Images from Text
- OpenAI Blog — Reducing Bias and Improving Safety in DALL·E 2
- OpenAI Blog — Lessons Learned on Language Model Safety and Misuse
- Engadget — Is DALL-E’s art borrowed or stolen?
- Forbes — AI Is Coming For Commercial Art Jobs. Can It Be Stopped?
- Forbes — Midjourney Founder David Holz On The Impact Of AI On Art, Imagination And The Creative Economy
- The Verge — Getty Images CEO says firms racing to sell AI art could be stepping into illegal territory
- The Verge — Shutterstock will start selling AI-generated stock imagery with help from OpenAI
- The Verge — The New York Times prohibits using its content to train AI models
- Jake Parker — The problem isn’t AI, it’s IA
- Seth Godin — Patterns, culture and theft
Videos
- Vox — The text-to-image revolution, explained
- Sam Does Arts — Why Artists are Fed Up with AI Art
- Proko — AI Ethics, Artists, and What You Can Do About It
- Palle Schmidt — Dave McKean on the Impact of AI for Artists
- Basement Picasso — AI Art – Biggest advance, or the greatest theft that artists will ever see??
- Tim Mcburnie – The Drawing Codex — Survive the AI Art Revolution
Podcasts
- How I Built This Lab — HIBT Lab! OpenAI: Sam Altman
- Hard Fork — Generative AI is Here. Who Should Control It?
- ComicLab — Special Guest Bill Barnes talks about A.I. comics
Data Protection
- Think Tank European Parliament — The impact of the General Data Protection Regulation (GDPR) on artificial intelligence
- Science — Legal reform to enhance global text and data mining research
- Social Media Today — LinkedIn Wins Latest Court Battle Against Data Scraping and the Misuse of User Information
- Hugh Stephens — Britain’s Proposed Approach to Text and Data Mining (TDM) for AI: How Not to do It (A Lesson for Canada and Others).
Copyright
- Which? — Who owns your online photos
- LegalZoom — What are the limitations of copyright protections?
- The Guardian — ‘Music is so different now’: Copyright laws need to change, says legal expert
- Farrer & Co — Status update on the proposed new copyright law exception to permit text and data mining for commercial use
- AI Business — AI-Generated Comic Book Could Lose Copyright Protection
Image Generative AI
Thanks Geo, a great and thoughtful overview of the dynamics between human and technology whereby the lines have become increasingly blurred since the industrial revolution and automation became a primary currency in human endeavor.
The only detail I could add to this, is the nuance that all machine learning / data models are currently based on very human content derived from our collective knowledge base. AI is a representation of how we think, feel, and create. What’s important to us and what we project into the world. Based on this fact it means that AI will not create anything new that hasn’t already been thought about, written about or even painted already by us humans… and then it merges such ideas and extrapolates them to create seemingly “new” solutions.
There appears to be a fear of AI overtaking humanity in many areas. These fears are warranted from an individualist lens. However, in terms of macroeconomics or general collective human endeavor, AI will not be able to surpass us. Much like the industrial revolution Human and AI (technology) will be a symbiotic relationship.
Until AI’s can begin modeling their data on their own resulting outputs or on other AI’s – AI will be limited to the work on the fringes of human endeavors and understanding. It will not go beyond this until AI truly begins to think for itself and has motivations to do so.
It’s an interesting time. Although AI’s development provides a huge leap forward for humanity, we should never underestimate human ingenuity and ability to forge new spaces and find new frontiers – with or without the help of AI, which you very nicely and eloquently highlighted in this article.
Cheers!
Hey Tarkan, Yes, data bias is a huge concern in AI and one of many hurdles it faces. I wanted to include it and other concerns in the article, but they started to pull the focus away from how AI will affect commercial art. So, thank you for raising this topic.
Some of the resources links take the time needed to explore the data bias.