AI Art: How artists are using and confronting machine learning | HOW TO SEE LIKE A MACHINE
TLDRThe video script delves into the intersection of AI and art, showcasing how artists are leveraging AI not just as a tool but as a medium to question and explore the nature of perception, creativity, and the existential divide between human and machine will. It highlights the generative turn in art, where AI's potential to dream and speculate beyond human-defined categories is examined, along with the ethical implications of AI's inherent biases and the potential for a more equitable society through innovative uses of technology.
Takeaways
- 🤖 AI is increasingly integrated into daily life, but there is a general lack of understanding about it.
- 🎨 Artists are using AI both as a creative tool and as a means to educate the public about AI's capabilities and implications.
- 🔮 Some artists are exploring AI's 'unsupervised learning' to push boundaries and challenge the conventional uses of technology.
- 🧠 The script discusses the concept of AI 'dreaming' and speculating about possibilities beyond human-defined categories.
- 🔍 AI's ability to learn and create is influenced by the data it's trained on, which can introduce biases and skewed perspectives.
- 🖼️ Refik Anadol's work at MoMA demonstrates how AI can reinterpret and transform large datasets into new forms of art.
- 🔑 The dialogue highlights the importance of questioning the labels and categories we assign to images and concepts in AI systems.
- 🌐 AI's generative capabilities are poised to revolutionize various creative fields, from illustration to film and publishing.
- 🔬 The script points out the inherent biases in AI training datasets and the real-world implications of these biases.
- 👁️ The perspective that AI offers is different from human perception, challenging our traditional value systems and points of reference.
- 🌐 The full impact of AI systems must consider their entire life cycle, from resource extraction to disposal, to understand their true cost to society.
Q & A
How do artists interact with AI in terms of understanding and confronting it?
-Artists are using AI both as a tool for creation and as a subject to explore and challenge. They intervene in AI processes to ponder existential questions about free will, human will, and machine will, as well as questions of perception, such as how to see things not made for human eyes.
What is the significance of artists using AI in ways it was not originally intended?
-By using AI in unconventional ways, artists are not rejecting technology outright but embracing it to make it do something else. This approach allows them to experiment, subvert, or divert the technology, pushing its boundaries and exploring new possibilities.
What breakthroughs in AI research were mentioned in the transcript?
-The transcript mentions breakthroughs such as OpenAI's DALLE, DALLE-2, ChatGPT, Stable Diffusion, and Midjourney. These are supervised, labeled, multi-model AI algorithms that facilitate interaction.
Can you explain the concept of 'supervised learning' in the context of AI?
-In supervised learning, humans tag information from the outset, providing examples like 'this is a pencil' for the AI to learn from. AI then becomes adept at creating or identifying objects based on these labeled examples.
What is the difference between supervised and unsupervised learning as discussed in the script?
-Supervised learning relies on human-provided labels for training data, whereas unsupervised learning allows the machine to find patterns and tag data on its own, often resulting in a 'black box' where the inner workings are not fully understood.
How did Refik Anadol use the MoMA archives in his artwork?
-Refik Anadol used the entire metadata of MoMA's archives, consisting of around 138,000 images, to create custom software artwork. This artwork interprets and transforms MoMA's collection data, presenting it in a large-scale, ever-changing presentation that explores potential AI dreams and imagination.
What is the significance of the 'latent space' mentioned by Refik Anadol?
-The latent space refers to the multi-dimensional representation of data in AI models. By navigating the latent space of MoMA's archive, Anadol's artwork reconstructs potential AI dreams, breaking down traditional categorical biases and exploring a convergence of past, present, and future.
What is the 'generative turn' mentioned by Kate Crawford, and why is it significant?
-The 'generative turn' refers to a pivotal moment where traditional understandings of creative fields like illustration, film, and publishing are rapidly changing due to AI's ability to generate new content. This shift challenges conventional notions of creativity and authorship.
How do AI systems reflect biases, and why is this a concern?
-AI systems are trained on datasets that can contain inherent biases, which are then reflected in their outputs. This is a concern because it can lead to skewed perspectives and reinforce stereotypes or discriminatory practices.
What is the role of artists in the conversation about AI and its societal impact?
-Artists bring a unique perspective to the conversation about AI, informed by thousands of years of thinking about the nature of images and meaning-making. They challenge the assumptions and biases in AI systems and explore the cultural and political implications of these technologies.
What is the 'anatomy of an AI system' as discussed by Kate Crawford and Trevor Paglen?
-The 'anatomy of an AI system' refers to understanding the full life cycle of an AI, from the extraction of rare earth minerals for its components to the end of life of the devices. This comprehensive view reveals the broader environmental and societal costs and implications of AI technology.
Outlines
🤖 Artistic Exploration of AI
This paragraph discusses the intersection of art and artificial intelligence (AI). Michelle Kuo and Paola Antonelli explore how artists are not just passive consumers of AI but are actively engaging with it to ask existential questions about free will and perception. They highlight how artists are using AI not as it is intended but to push its boundaries and explore new possibilities. Refik Anadol mentions breakthroughs in AI research and how conventional supervised learning is being challenged by unsupervised learning, where AI creates its own classifications and dreams up new realities. The conversation includes the use of MoMA's archives to create a dynamic, ever-changing artwork that represents AI's unique perspective on art.
🔮 The Generative Turn and AI's Biases
The second paragraph delves into the generative capabilities of AI and the biases inherent in its systems. Kate Crawford and Trevor Paglen discuss the 'generative turn,' a pivotal moment where traditional creative processes are rapidly changing due to AI. They challenge the notion of AI's objectivity, pointing out that AI systems are biased from the start by the data they are trained on. Paglen's work, 'Behold these Glorious Times!', is highlighted as an example of how artists reveal the biases in AI training data sets. Crawford emphasizes the oversimplification of complex realities in AI labeling and the risks of a 'bleached' version of the world, while Paglen discusses the importance of artists' historical and cultural perspectives in understanding images, which contrasts with the engineering computer science tradition.
🏭 The Historical Dance Between Art and Technology
This paragraph examines the historical relationship between art, technology, and machines. Paola Antonelli and Michelle Kuo reflect on how artists have long been fascinated by and responded to technological advancements, from the industrial revolution to the current AI era. They mention Marcel Duchamp's readymades and the Machine Art exhibition at MoMA, illustrating the evolving appreciation for machines and their integration into art. Antonelli discusses the evolution from machines needing to understand human design to humans making machines' concepts more understandable. Trevor Paglen and Kate Crawford address concerns about AI's role in society, including the potential for wealth and power consolidation, and the environmental and societal costs of AI systems. They advocate for radical uses of AI that challenge expectations and possibly work against the systems themselves, hinting at a future where AI's role in creativity and reality definition will be increasingly questioned.
Mindmap
Keywords
💡AI
💡Supervised Learning
💡Unsupervised Learning
💡Machine Will
💡Perception
💡Existential Questions
💡Generative Turn
💡Algorithmic Bias
💡Latent Space
💡Collective Consciousness
💡Inequitable Society
Highlights
AI is integrated into daily life, yet there's a lack of understanding about it.
Artists are using AI as a tool to explore existential questions of free will and perception.
Some artists aim to intervene in AI processes to provoke thought on human and machine will.
Artists are redefining how to use AI, pushing it beyond its intended functions.
Recent AI breakthroughs like DALLE-2 and ChatGPT are reshaping interaction with technology.
In supervised learning, humans tag information, directing AI's creative output.
Unsupervised learning allows AI to create its own classifications and dream-like imagery.
The MoMA exhibition 'Unsupervised' explores AI's speculative imagination beyond human labeling.
AI's complex classification system creates a galaxy of data with potential for new existences.
AI-generated dreams challenge traditional categories and biases, offering a multi-dimensional perspective.
The generative turn in AI is rapidly changing traditional creative processes in various industries.
AI systems are not objective; they reflect the biases present in their training data.
Trevor Paglen's work exposes the cultural and political biases embedded in AI systems.
AI's potential for simplifying and bleaching the world's complexity is concerning.
Artists bring a unique perspective on the meaning and implications of AI-generated images.
The history of art and technology shows an evolving relationship between humans and machines.
Designers have been examining the role of machines in art since the early 20th century.
AI's potential for solving hard problems is overshadowed by concerns of wealth and power consolidation.
The full life cycle of an AI system, from mining to disposal, has a significant planetary cost.
AI algorithms may redefine creativity and challenge our understanding of reality.