creativity – OPEN

AI idea-finding: thoughts dancing to a new rhythm

Stop pushing buttons. Start making moves.

Loose strands forming the shape of a dancer making stretching and flowing movements. — Excerpt from *Fascia*, an artwork by Tobias Gremmler; dance by Lico Kehua Li.

We can’t control systems or figure them out. But we can dance with them!
—Donella Meadows, pioneering systems thinker and environmental scientist.

Currently, many people are making and sharing recipes for things you can make with AI. Rather than say recipes, you have to put on a serious face and call the play experimentation and the recipes have to be called prompt engineering, in a voice two octaves lower than usual.

I’ve been doing this too and sharing some in Antonym, like making a ordered shopping list or whole presentations and writing up meeting notes.

Recently, I had an experience that made those things feel like basic moves, like practice; suddenly, I’d learned how to strong them together to into dancing. It started with a simple experiment and then spiralled into an adventure with ideas. Once it was happening, I almost couldn’t stop – a sensation like when you start to fly in a dream, or realise you are piloting some strange and wonderful vehicle. A little out of control, but not the red shoes of the fairy tale.

Accelerated inspiration

The following isn’t a recipe, but it does include some of the steps I took, and some of them may be useful to you. I was using ChatGPT for the most part. If you would rather not pay the extra $20 to use this, you may get similar – or “good enough – results with the free standard ChatGPT or its rival Anthropic’s Claude. Apps such as Poe will let you have a few free goes on both platforms. But the tech isn’t the point here; the way of thinking and the possibilities and how it feels are what I want to share.

It started something that sparked some curiosity. My wife, whose passion is garden design, was talking about a famous designer called Piet Oudolf. He creates gardens that are self-sustaining, support wildlife and other wonderful things. The one I knew was the High Line garden in New York, a long, narrow strip of a park made on an old elevated subway track in the Meatpacking District.

The High Line, New York City (20th Street, looking downtown). Credit: (cc) Beyond My Ken.

His planting schemes often look like wild prairies but are meticulously planned systems of systems. The drawings, she said, look unlike any other type of garden design, more like scientific diagrams of mitochondria than planting plans. Here’s an example:

And here’s some detail from a design which expresses the way that different groups of plants interact with each other over the seasons.

Image: Segment of planting design for Hauser & Wirth Somerset garden Credit: (cc) John Lord

The description made me think about systems thinking of course; Oudolf seems like a systems thinker?—?focused on connections and how different elements interact with each other.

Exploring a connection

I asked the latest version of ChatGPT (currently premium, running GPT-4) this: Can you show me some images of Piet Oudolf’s designs and tell me about his design philosophy?. It returned the following, a useful primer on his approach which tallied with and built on what I knew already:

Piet Oudolf is a Dutch garden designer, plantsman, and writer who is renowned for his innovative approach to planting design. He has been a leading figure in the New Perennial Movement, a gardening style that emphasises the use of perennial plants, particularly grasses, and focuses on creating naturalistic landscapes. His approach to garden design is characterised by several key principles:

– Plant-driven design: Oudolf places a strong emphasis on selecting plants for their aesthetic qualities, such as form, texture, and colour, rather than focusing solely on their horticultural characteristics. This plant-driven approach results in gardens that are visually engaging and have a naturalistic feel.
– Seasonal interest: Oudolf’s designs prioritise plants that provide interest throughout the year, not just during the peak of the growing season. He selects plants that have beautiful seed heads, attractive winter silhouettes, or striking fall colour, ensuring that the garden remains engaging even during the colder months.
– Layering: Oudolf’s gardens are characterised by a layered planting approach, in which plants are arranged in a way that mimics the structure of natural ecosystems. This results in gardens with a diverse range of plant heights, densities, and textures, creating a sense of depth and complexity.
– Ecological considerations: In his designs, Oudolf often incorporates native plants and emphasises the importance of supporting local ecosystems and wildlife. He believes that a garden should not only be aesthetically pleasing but also contribute to environmental conservation and biodiversity.
– Embracing imperfection: Oudolf’s gardens are not meticulously manicured spaces, but rather, they embrace the natural processes of growth, decay, and regeneration. This results in landscapes that have an organic, evolving character, which reflects the inherent beauty of nature.

Some of Oudolf’s most famous projects include the High Line in New York City, the Lurie Garden in Chicago, and the Millennium Garden at Pensthorpe Natural Park in the UK. His work has had a profound impact on contemporary garden design, inspiring countless designers and gardeners to embrace a more naturalistic and ecological approach to landscaping

Amazing. Was there much writing about the connections with systems thinking in his approach? ChatGPT said no, but that it did see parallels:

While there isn’t a direct and widely-discussed connection between systems thinking and Piet Oudolf’s work in garden design, one can still draw some parallels between the two. Systems thinking is an approach to problem-solving that focuses on understanding the interrelationships and interdependencies within a system, rather than just focusing on individual components in isolation. It seeks to understand how elements within a system work together and how changes in one part of the system might affect other parts.

In the context of Piet Oudolf’s work, one could argue that his approach to garden design embodies some principles of systems thinking, such as:
– Emphasis on relationships: Oudolf’s designs prioritise the relationships between plants, considering how they work together in terms of form, texture, colour, and seasonal interest. By focusing on these relationships, he creates gardens that have a more cohesive and naturalistic appearance.
– Holistic approach: Oudolf considers the garden as a whole, rather than just focusing on individual plants or elements. His designs take into account the overall structure, composition, and aesthetic qualities of the space, as well as its ecological impact and support for local wildlife.
– Adaptation and change: Oudolf’s gardens embrace the natural processes of growth, decay, and regeneration, allowing the landscape to evolve and change over time. This approach is consistent with systems thinking’s emphasis on understanding the dynamic nature of systems and their ability to adapt to changing conditions.
– Ecological considerations: Oudolf’s emphasis on native plants and supporting local ecosystems is in line with systems thinking’s focus on understanding the broader context in which a system operates, as well as its connections with other systems. By considering the ecological impact of his designs, Oudolf contributes to the overall health and sustainability of the environment.
While these parallels can be drawn, it is important to note that the connection between systems thinking and Piet Oudolf’s work in garden design is not explicit or widely recognised. However, it can be interesting to consider how his approach to design might resonate with the principles of systems thinking and contribute to a more holistic and interconnected understanding of garden design and landscape architecture

Amazement aside, what was going on here? This was better than search engines. Not just the creaky SEO-muddied search results you get from searching for anything, but s going beyond the question and finding synthesis, connections between fields. It was the kind of short article or blog post I might write over a couple of hours, once I’d been able to read enough about Oudolf and consult my notes on systems thinking concepts. Instead, it was right here in front of me. This is — or something similar – is what I would have written.

Combinatorial innovations a–gogo

I felt slightly giddy.

What did I want to do now? Turn those notes into an article? Or go further?

Oh, further?—?of course, go further. Where I was headed was how I might apply some of this as metaphor or inspiration for my own work. What would Oudolf’s approach and lessons from design thinking apply for, say something like designing an organisation, the way that teams or departments fit together. This was something I had found hard in the past and suspected there could be better ways of thinking about it. One stumbling block is that organisation design is not an org chart; but, org charts are useful for some things, and expected by people. As soon as you produce a chart, that’s what humans’ territory and hierarchy-attuned minds go to, and it’s hard to get everyone thinking about other aspects, like how the bits work together.

So, I asked:

Piet Oudolf’s designs are famous for how they express his ideas in diagrammatic form. What might an organisation design visualisation look like if you took that approach – describe the images and components and then again as prompts to create those that could be used with DALL-E 2 or Midjourney to create useful elements and diagrams.

And got another analysis with about eight bulleted paragraphs explaining connections such as:

Emphasis on relationships and interconnectedness: Just like Oudolf’s approach to garden design, organisational design should focus on fostering strong relationships and understanding the interconnectedness of various departments and roles. Designing an organisation with a focus on collaboration and communication can create a more cohesive and efficient system.

Here’s the full response:

Drawing inspiration from systems thinking and Piet Oudolf’s work in garden design can lead to several ideas and principles that might be applicable to organisation design. Here are some concepts that could potentially contribute to better ways of designing organisations:

– Emphasis on relationships and interconnectedness: Just like Oudolf’s approach to garden design, organisational design should focus on fostering strong relationships and understanding the interconnectedness of various departments and roles. Designing an organisation with a focus on collaboration and communication can create a more cohesive and efficient system.
– Holistic approach: In line with systems thinking, organisation design should take a holistic view, considering the organisation as a whole rather than focusing solely on individual components or departments. This includes evaluating the organisation’s culture, values, and objectives, and aligning them to create a unified and purpose-driven entity.
– Adaptability and flexibility: Organisations, like gardens, need to be adaptable and flexible to thrive in a constantly changing environment. Organisational design should incorporate structures and processes that allow for adaptation and responsiveness to external and internal changes, facilitating innovation and continuous improvement.
– Diversity and layering: Similar to Oudolf’s layered planting approach, organisations can benefit from a diverse workforce with a range of skills, experiences, and perspectives. This diversity can foster creativity, innovation, and resilience, and contribute to the overall success and sustainability of the organisation.
– Ecological considerations and sustainability: organisations should consider the broader context in which they operate, including their social and environmental impacts. Designing organisations with a focus on sustainability, ethical practices, and social responsibility can contribute to a healthier and more resilient system.
– Embracing imperfection and learning from failures: Like Oudolf’s gardens, organisations should recognise that imperfection is an inherent part of growth and progress. By fostering a culture that embraces experimentation and learns from failures, organisations can continually evolve and improve.
– Balancing autonomy and interdependence: In a well-designed garden, plants thrive both independently and as part of a larger ecosystem. Similarly, organisations should strike a balance between providing employees with autonomy and fostering a sense of interdependence, encouraging individuals to contribute to the organisations overall success.

By applying these principles and ideas, organisations can draw inspiration from systems thinking and Piet Oudolf’s work in garden design, potentially leading to more effective, resilient, and sustainable organisational structures.

Now show us what it looks like

Then there were the prompts. I tried this one in the DALL-E 2 and Midjourney image generators. DALL-E 2 was disappointing. Midjourney gave a credible and interesting set of images:

I added an instruction to make them appear hand-drawn, like Oudolf’s:

Exactly what I was imagining. Better, maybe.

So, in about half an hour I’d learned huge amounts, explored an untapped area of adjacent innovation in organisation design and mocked up how we might re-imagine org charts.

I was over-stimulated. Excited by the experience. All the dopamine and other hormones fired by my brain sensing new knowledge had me feeling almost dizzy. I had to slow down.

What does this experience tell us about how we can work with AI?

Far more exciting than having AI create content is trying out ideas. When my colleagues in graphic design or creative use image generation AI tool often it if for the mood boards and rough creative territories that would have taken hours or days to produce as part of the process in the past. Writers and knowledge workers can do the same, but aren’t necessarily used to working with multiple possible versions of the same work that are very different from one another.

AI can be a smart study partner or tutor. This was a learning experience as much as a writing or research process. Taking existing knowledge into an adjacent domain is always exciting, what made this different was how fast you could get from question, to idea, to concept to exploration.

The better your questions, the further you go. The chat is more than a simulation, it is a conversation with texts and networks of knowledge. Domain knowledge and curiosity can take you in to other fields fast.

Search engines are so screwed once this catches on. This was a specialised, nerdy, almost academic enquiry, but once you experienced the efficiency and felt the potential of thinking like this, each time I turn to a traditional search engine it feels very limited. Bing has some of this quality, not least because it uses OpenAI’s software. Google could transform this approach.

Reasonably priced, for now. Short-term, the $20 a month for ChatGPT premium seems like a bargain for someone who will use it in their work. Other tools like Claude from Anthropic are similar, and the note-taking app Notion blends Claude and ChatGPT, I believe.

Fancy a go?

The following would be my advice to a colleague or friend who is a knowledge worker or creative and is comfortable, but not expert with using search engines and standard productivity tools (Microsoft Office, Google Worksuite or equivalent).

Invest $20 a month and an hour a day to learn to dance with AI. I recommend a month at least with ChatGPT, Notion, or Claude (all of which charge a similar amount) and using it on at least one task where you have a little time to experiment. I’ve found it saves more time than it takes to learn on things like planning, analysis and admin (meeting notes, preparing structures and guides for writing anything from business documents to copy), but this may not be the case for everyone.
Keep notes. Take screen grabs and write a little journal of what you notice, what works and what doesn’t. These may be useful to review later on, but even if you don’t, it will make sure you reflect, if only briefly in the moment, on what is happening.
Be disciplined about process. This is a topic for another article, but being clear about the steps you will take to create something, even designing a big meeting, or sending a concise analysis of a topic, and creating a check-list for the job will help you see where it will be worth trying working with a generative AI tool.
Double check outputs for facts and accuracy. GPT-4 and equivalent generative AIs are more accurate, but still liable to make things up?—?this includes checking links are real and references to articles or other sources actually exist and say what it claims they do.
Find your rhythm, make your own moves. Every breathless claim of a prompt or tool that can save HOURS or make $$$$$?—?there are so many on social media currently?—?should be treated with extreme caution. What worked for one person may not replicate for you. Even scientists find it hard to replicate results in some controlled experiments. Instead of looking for magic spells or artefacts, use the metaphor of dance to learn how to give and take with the tools, find what’s possible and what suits your mind. As well as the unpredictable nature of AI sometimes, the different demands of our specific work and our individual brains are all going to require different techniques and use of tools.

In the spirit of that last point, let’s end with a fuller version of the Donella Meadows quote we started with:

We can’t control systems or figure them out. But we can dance with them!

I […] learned about dancing with great powers from whitewater kayaking, from gardening, from playing music, from skiing. All those endeavours require one to stay wide awake, pay close attention, participate flat out, and respond to feedback. It had never occurred to me that those same requirements might apply to intellectual work, to management, to government, to getting along with people.

Let’s dance.

This post was developed from a seed article on Antonym, my newsletter. To subscribe please visit Antonym on Substack.

Outliner thinking

In this MIT Technology Review article about different writing and blog authoring tools – As We May Type – Paul Ford describes a tribe I wasn’t aware existed, but once described I knew immediately I was a part of – “outliner people”.

Outliners were one of the first writing tools available on computers and they continue to be very important. Ford defines it as…

…a kind of mental tree. Say level 1 is a line of text. Then level 1.1 would be subordinate to 1, and 1.1.1 subordinate to 1.1; 1.2, like 1.1, is subordinate to the first line. And so forth.

Personally, I use Omnioutliner Pro, CarbonFin’s excellent Outliner app for IOS, as well the outlining functions in Evernote and Curio on occasion. I picked up the practice from Jim Byford and my now business-partner Jason Ryan, who conjures major projects, intricate strategies and complex plans on a screen, turning an interesting conversation into an action plan and the beginning of a briefing document or proposal.

I like mindmaps, but outliners suit my needs more often. Sometimes an idea will be developed in a mindmap and then be transferred (as an OPML file – Curio does this automatically very well) to an outline and later that outline will turn into a Google Doc, Pages or Word file that can be made more beautiful and complicated and ready for sharing with the world outside the project team.

It’s a case of the right tool for the way you need to think in a given situation. But also, the right tool in chain of tools that can become a workflow that means you move from idea, to concept, to model, to prototype to plan in smooth transitions, with as little friction and cognitive costs between each step as possible.

More on that thought in the next post…

Things I learned from Ed Catmull

A few months ago I heard Ed Catmull, president of Walt Disney Animation Studios and Pixar, give a talk about leading creative organisations.

Apart from his obvious experience and track record of success, what was clear was that he had thought very deeply about some crucial questions about leadership.

These are some notes about what he said and thoughts that he provoked. To be clear – they are not direct quotes – they are my recollections and thoughts based on my notes of his talk (what I learned rather than what I heard).

Be prepared for near death experiences on projects. All Pixar movies “suck” at first. They are radically altered again and again until they work. Every Pixar film except one – Toy Story 3 – has gone through a phase of intense crisis during its development.

Most people want to avoid the “near death” part of the creative experience, but it is very often essential in order to get to something really good. (This reminded me once again of the valley of creative despair that is the liminal state.)

A leader’s job is to maintain a balance of power. In a studio – just like an agency – there are business functions like finance, production, creative, marketing, technology etc. Organisations fail when one of these functions “wins” and dominates the others with its agenda.

For instance, in studios where production wins, films are produced on time and on budget, but creatives become demoralised and produce lower quality work and talent leaves. A CEO or President needs to make sure that no part of the organisation becomes dominant and skews resources with its particular agenda.

Business books are curiously free of content. Very often business books state obvious truths and avoid more difficult questions.

Smart people make stupid decisions. This is a puzzle that more business books could do with taking on – rather than succumbing to narrative bias, or focusing on successes. There should be more books about failure, more conversations about why we do stupid things.

Leaders can’t see the things that are going wrong. When an organisation is bound for failure What are the forces that I can’t see, is the question a leader needs to constantly ask themselves. People will be behaving badly at times – but they will never do it in front of you.

You need to make the information flow separate from the organisational structure. This reminded me of Churchill, who set up the Office for National Statistics so he could hear what was really going on – rather than allowing each department to gather data and report in their own way, influenced by their various agendas.

You need people to be candid, not just honest. Politeness, respect, embarrassment, fear, blinkered-visions/solipsism, and other things can stop people from being candid. His job as CEO is to spot those things and stop them. Often leaders will prevent candour with their presence in a room, unless they build trust and make it clear what behaviours are acceptable.

Protect new ideas. New ideas are vulnerable, delicate things. If they are good ideas they need to be protected, allowed to develop in a safe space. Success at Pixar and other creative companies is about creating safe spaces for creatives and ideas to flourish.

We are always operating in a fuzzy space. We have to be comfortable doing that. Again, invoking the liminal state for me, Dr Catmull talked about the need for creative leaders and creatives to work in and with uncertainty. We cannot deliver genius on schedule, we need to be comfortable with that. We do not know what the final iteration of the story will look like, we have to be comfortable with that. We don’t know what the technology or media landscape will look like more than about six months from now, but we have to make plans for the next six years – and know they may look as different in the final version as Up looked from the first idea for that story (a castle floating in the clouds full of people at war with the people on the ground, apparently)…

Artefact Cards and liminal states: creative thinking breakthrough tools

There’s a long, long list in my subconscious that I hardly dare look at: Things I Should Have Blogged.

The items comprise three types:

Important ideas that have taken up residence in my head. For instance, liminal states.
Useful tools and ways of working. For instance Artefact Cards.
Opinions taking shape. For instance, just because a system like digital advertising is corrupted doesn’t mean it won’t be with us for decades to come.

I may come back to the third and pot it out in the nursery of ideas here on this blog, with a media agency-proof fence around it to give it a fair chance of developing or not, but for now I’ve got a chance to right the first two examples in one post.

This week, I had a lovely conversation with John Willshire, who developed the Artefact Cards product, about how I have been working with them. You can listen to the whole thing here as John recorded it with a very snazzy microphone and iPad Mini set-up.

Artefact Cards are a really simple tool. Playing card size bits of card, white on one side and coloured on the other. You draw words and pictures on them with a Sharpie pen and them lay them out, re-arrange them and in this way organise thoughts and ideas.

As we talked, we got onto the subject of the liminal state in the creative thinking process (which for my money includes developing strategy). My friend Jim Byford introduced me to this immensely useful concept.

In the context of creative and strategic thinking, the liminal state is what you find yourself in just before you have a breakthrough, or just before you fully understand something, make it yours. For instance, if you can recall trying to learn your lines for a play, the liminal state is where you are just before the words settle and take up residence in your memory – and then you can start using them, adding your inflections and emotions, making them your own.

I’ve very often found the thinking at this point in the creative process intensely uncomfortable. Whether writing a book or a plan or a pitch – it’s a kind of temporary agony, a dark tunnel I pass through where I think you know nothing and will never have another good idea again, and then it passes and there’s the the idea I need, the answer that fits.

Knowing that this is something called a “liminal state”, it makes it easier to handle. In psychology / neuroscience, this is an example of “affect labelling“. If you can name the feeling you have, you can put yourself slightly outside it, understand what is happening to you and that it will pass.

The other thing that understanding the liminal state does is help you to stop trying to “jump to the answer”, as Jim put it to me. Because liminality feels uncomfortable, you want out – to end the feeling and go with the first idea, the obvious one, the easy one. The danger here is that your creative/strategic solution will be mundane, run-of-the-mill and doomed.

You have to go through the confusion, live with it for a little while, sit still while the ideas and thoughts, disconnected and jagged, whiz around your head.

Then they settle. Then you see it: what it is all about.

It’s simple, it was there all along… as Duncan Watts points out, it feels obvious once you it is something that you understand. You pitch it to yourself: it works. You pitch it to a colleague: they don’t hate it, maybe even like it. With each airing the idea gains coherence, legitimacy – becomes more eloquently and credibly articulated as you and others breathe belief into the thing.

Speaking with John Willshire about how I had been using his Artefact Cards, I realised that I like them because they are a good tool for helping that settling process, of working steadily through the seemingly nonsensical maze of thoughts, ideas and concepts and helping some kind of order emerge. Much like throwing down ideas on a white-board, scribbling out mind-maps or any other visual thinking method – but they feel slightly more agile – you can move ideas around, try them in different shapes more rapidly.

In the example I talk about, it’s not even that I reached the solution – the outline of an ebook in this case, but I was able to move on to that only after I had made sense of all of the ideas. Seen their shape laid out in this way. That’s something John says is a recurring theme in people’s use of the cards – seeing the “shape of ideas”.

Artefact Cards are another tool in the box for thinking, perfect sometimes for working through those liminal states. Worth a spin with the trial pack, I reckon.