Pollinator Pathmaker is an algorithmically designed planting scheme that exists as physical gardens — its artist calls them editions — optimised not for human viewers but for pollinating insects. On 22 October 2026, designboom published an interview with artist Alexandra Daisy Ginsberg and LAS Art Foundation curator Sophie Korschildgen about the Berlin edition, which has been growing in the forecourt of the Museum für Naturkunde since June 2023.
If you are in Berlin, note the date first: LAS’s own What’s On listing gives the edition’s run as 20 June 2023 – 1 November 2026. That is ten days away. We would not read it as the garden being dug up — this is a living artwork with a maintenance team and a research programme attached — but the foundation’s public listing for it closes at the start of November, so if you were going to go and look at it as a commissioned artwork rather than a flowerbed, go now. It is free and open to the public in the forecourt.
The scale, from LAS’s own microsite: more than 7,000 plants across a 722-square-metre plot, designed by what the foundation calls a custom-built algorithm.
Two things in the interview are worth the attention of anyone working at the intersection of code and the physical world. One is a number. The other is a correction.
The number
From Korschildgen, describing the monitoring work:
In Berlin, especially, we have a wild bee specialist who analyzes the different wild bee species that come to this garden… The average number in areas of the city that are called ‘pollinator-friendly’ is about 30 to 50 different wild bee species in one place. At the LAS edition last year, there were already 83 different species — quite a bit above the average!
Wild bees are the measurement target because they are an indicator species: if wild bees are doing well, other pollinator populations likely are too.
Three caveats worth stating plainly, none of which the interview hides. This is a species count at one site, not a controlled comparison. The figure is from last year. And the plot sits in front of a museum that over a million people visit annually, which is not a typical urban green space. Formal studies are running — one at the Eden Project in Cornwall, one on the Berlin edition, and a UK-funded study with the University of Exeter and the University of Edinburgh in a Cornish village, which Ginsberg won a research grant for two years ago. The published results are what will settle whether the planting algorithm beats conventional pollinator-friendly planting. But 83 against 30–50 is the first hard number this project has put in public, and it points the right way.
The correction
This is the line to keep:
Alexandra Daisy Ginsberg’s project Pollinator Pathmaker was such a beautiful concept in that sense because it’s a global series of gardens that are designed by an algorithm (not an AI, that’s important) to benefit pollinating insects.
That parenthesis is the curator volunteering a distinction that almost every press account of algorithmic art now collapses. Pollinator Pathmaker runs a planting optimiser over a region-specific palette of plants, chosen from research by Ginsberg with pollinator and horticultural experts for that biogeographic region. It generates compositions that prioritise pollinator benefit. It is a solver with a defined objective function and expert-curated inputs. It is not a model that learned anything, and the people who made it would rather you did not say it was.
We have spent this month finding the opposite everywhere — a dev kit covered as an AI platform by everyone except its own vendor, among others. A curator pre-emptively narrowing the claim is rare enough to note.
How it works, and what you can run yourself
At pollinator.art the project opens with a question: “If pollinators designed gardens, what would humans see?” You set a location, a garden size and other parameters, and the tool generates an artwork — a planting composition drawn from the plant palette available for your region. As new commissions arrive, new region-specific palettes are added.
Ginsberg’s own description of the premise is the cleanest statement of it:
Pollinator Pathmaker began with a question: what happens when we create an artwork for other species rather than ourselves?
The project was originally commissioned for the Eden Project in Cornwall. LAS was the first international commissioner and partner, and adapted the system to Berlin — specifically, Korschildgen notes, to the continental biogeographic region, because the UK and Germany are not the same growing conditions. Large-scale plots also exist at the Serpentine in London, and a 2025 edition at Google Gradient Canopy in Mountain View, California. Alongside the gardens there are DIY editions anyone can generate and plant, plus print, tapestry and video works.
This year’s landmark
we reached our goal of being in every district of Berlin with a small edition—there are 20 DIY editions across the city, and 21 editions including the one in front of the Natural History Museum.
That is a distributed artwork completing coverage of a city, one volunteer-planted plot at a time.
One small inconsistency, for the record: LAS’s own microsite describes the campaign as having brought Pollinator Pathmaker into “almost all districts” of Berlin, while the curator told designboom they had reached the goal of every district. The gap is probably just a page that has not been updated since the last planting, and twenty volunteer-planted editions either way is the achievement. But the microsite and the interview do not agree, and we would rather say so than pick the better-sounding one.
Korschildgen’s favourite moment to see it is now:
if you were to go, you would see asters across the whole garden. It’s a type called ‘October Skies,’ which means it blooms lilac in October. Right now, in autumn, is the time when you see this almost algorithmic arrangement and formal strangeness of this artwork.
The maintenance problem is the artwork
The part that will resonate with anyone who has shipped an installation: the piece is maintenance-heavy, and maintaining it means enforcing the generated plan against reality.
It’s artificial nature for nature
— Ginsberg’s phrase, as Korschildgen relays it, stressing that this is an artificial arrangement. Not a garden, and not plants that would naturally grow together at that density under those conditions. So the team weeds out volunteers even when a weed would benefit pollinators, because the planting plan is the work. And since the plot is full of benches outside a busy museum, it accumulates interventions:
And also, rogue tomatoes! For a couple of years, we had rogue tomato plants that were growing because perhaps people were eating their lunch and throwing the bad tomatoes into the artwork.
A generative work whose output is alive, whose substrate keeps editing itself, and whose conservation involves deciding which emergent additions are vandalism and which are lunch. There is no version of that problem in software.
LAS’s own interest is stated as a research strand on interspecies futures, set up with the Natural History Museum and other partners, asking: how can planetary computational technology be developed to benefit the natural world rather than extract from it?
The honest answer so far is one plot, 83 species, and a study still to report. That is a better position than most work in this area can claim.
Sources
- pollinator pathmaker brings buzzing bees back to berlin with foliage-filled artworks — designboom
- Pollinator Pathmaker
- Pollinator Pathmaker Berlin — LAS Art Foundation
- LAS Art Foundation
- Alexandra Daisy Ginsberg: Pollinator Pathmaker, short documentary — LAS Art Foundation on YouTube
- Museum für Naturkunde Berlin