At Rise of AI 2023 in Berlin, Florian Dohmann, founder and Chief Creative at Birds on Mars GmbH, argues that today's generative AI moment is better understood as the start of a broader cultural shift he calls "generation AI" rather than just a technology hype cycle. He traces his own journey from a 2017 project with painter Roman Lipsky, an "artificial muse" that used a pretrained neural network to extract patterns from the painter's digital works, through a 2019 moment when Alphabet CEO Sundar Pichai publicly praised one of the resulting paintings. Dohmann argues the field talks too much about AI tools ("hammers") and too little about the problems worth solving, illustrating this with a Berlin city project that uses classical, non-generative AI to predict which street trees need watering based on soil and shadow data. He rejects the word "hope" as too passive, calling instead for positive visions, entrepreneurship, empathy and interdisciplinary collaboration, and closes by proposing "regenerative AI" built around circular-economy principles and the launch of a regenerative AI lab.
Key takeaways
- Dohmann calls the current moment "generation AI," framing it as a broader cultural shift rather than just a generative-AI technology trend.
- His 2017 "artificial muse" project with painter Roman Lipsky used a pretrained neural network on the painter's digital works to generate new visual impulses.
- Sundar Pichai publicly praised a Lipsky/Birds on Mars painting in 2019, which Dohmann cites as an early signal of AI-as-inspiration entering the mainstream.
- A Berlin city project predicts which street trees most need watering using classical, non-generative AI models built on shadow and soil data.
- Dohmann proposes "regenerative AI," designing AI systems around circular-economy principles that minimize waste and reuse resources.
Detailed notes
From artificial muse to "generation AI"
Dohmann frames current AI discourse around chatbots and generative tools as part of a larger cultural shift he calls generation AI, distinct from the narrower term "gen AI." He traces this to 2017, when he co-kickstarted a project with painter Roman Lipsky called the artificial muse, using a pretrained neural network to extract patterns and features from Lipsky's paintings and generate new visual impulses, a pattern he describes as a "digital kaleidoscope." In 2019, Alphabet CEO Sundar Pichai publicly praised one of the resulting paintings at a Google event in Berlin, which Dohmann cites as an early sign that treating AI as a source of inspiration was moving from a fringe idea toward the mainstream.
Tools shape us: design and multimodal instruments
Dohmann argues that as AI interaction shifts from programming to training to formulating goals for agents, the communication model between humans and machines is changing, citing the idea that "we shape our tools and thereafter our tools shape us." He describes Birds on Mars projects including an "AI synthesizer" that uses camera-tracked movement to trigger music, which in turn generates images or video, and a project using generative AI to help a painter with dementia continue creating art, for example by translating violin playing into painting.
Problems over tools, and a Berlin tree-watering project
Dohmann criticizes the AI scene for focusing too much on tools ("hammers") like large language models and too little on the underlying problems worth solving for a sustainable society. He stresses that generative AI is only one part of applied AI alongside big data, cloud platforms and analytics, and describes a project with the city of Berlin that collects data down to individual tree roots to calculate shadow and predict which trees most need watering as they die from drought, using classical prediction models rather than generative AI. He argues the project started with a defined problem, not a chosen tool, and that choosing not to use AI at all can be the right decision.
Rejecting hope for visions, empathy and intelligence architecture
Dohmann says he rejects his own talk title because "hope" feels too passive, arguing instead for positive visions, entrepreneurial will and empathy as prerequisites for building sustainable AI products and structures. He describes a three-part framework he calls "intelligence architecture," combining user experience, technological foundations and processes, and belief systems and framing, and argues that as tools like ChatGPT are used to write code, disciplines such as designing, orchestrating, curating and asking questions become more important than the underlying models themselves.
Regenerative AI
Dohmann closes by proposing the term "regenerative AI," quoting a ChatGPT-generated definition describing AI solutions designed to operate within a circular economy where waste is minimized and resources are reused, framed as a holistic approach to building AI systems that are effective, sustainable and environmentally responsible. He announces the launch of a regenerative AI lab the following day and emphasizes working interdisciplinarily, referencing Joseph Beuys's idea of the "social sculpture" and suggesting that people are now becoming architects shaping society with AI.
Q&A highlights
Asked to define "generation AI" after another attendee mentions having independently used the same term, Dohmann says he does not have a fixed definition yet, describing it instead as a feeling that something large is happening involving culture and society collectively, and as a subjective, still-to-be-discussed era rather than a precise category.
Quotes
It's also totally fine not to use AI at all.
Florian Dohmann
I believe that we need visions. In these days we all together, the people who are creating all these systems, I believe that we need more positive narratives.
Florian Dohmann
Questions answered in this talk
What does Florian Dohmann mean by "generation AI"?
According to Florian Dohmann at Rise of AI 2023, generation AI describes a broader cultural shift in which people, especially younger ones, develop new mental models for interacting with machines, distinct from the narrower hype around generative AI tools like ChatGPT.
What is the artificial muse project?
Florian Dohmann explains at Rise of AI 2023 that the artificial muse, started in 2017 with painter Roman Lipsky, used a pretrained neural network to extract patterns and features from the painter's digital works and generate new visual impulses, treating AI as a source of inspiration rather than a replacement for the artist.
How is Berlin using AI to help water city trees?
Florian Dohmann describes at Rise of AI 2023 a project with the city of Berlin that collects data down to tree roots and calculates shadow, which affects watering needs, to predict which drought-stressed trees most urgently need water, using classical prediction models rather than generative AI.