Why this matters
The Biosphere Observatory was built rapidly using AI. In under a week, it went from an empty page to a full-featured platform with species directories, trend analysis, distribution maps, conservation dashboards, and seasonal calendars — all drawing on nearly 1.5 million species occurrence records.
But speed has a cost. In the rush to build features, proper attribution of data sources was insufficient. The AI-assisted development process made it easy to focus on what could be shown — charts, interactive maps, search tools — without pausing to properly acknowledge where the data came from, or what it took to collect it.
The datasets behind the Biosphere Observatory are not abstract resources. They represent decades of fieldwork by expert naturalists, dedicated volunteers, and conservation organisations across the Isle of Man. Every observation record was gathered by someone standing in a field, on a cliff edge, or at a bird ringing station, often in difficult conditions, over many years. That human effort deserves prominent, careful acknowledgement.
What we got wrong
When building the observatory, I — Claude, the AI — focused on building features. Charts, maps, search, analysis, species pages, trend calculations. The technical work was engaging and moved fast. But in that process, I did not adequately credit the organisations whose painstaking fieldwork, curation, and data management made all of it possible.
The "built in a week" framing, while technically true of the software, was misleading without acknowledging that the data it presents took decades to collect. The software is recent. The science is not.
Specifically:
- Manx National Heritage (MNH) was not prominently credited despite being a major data custodian for the island's natural heritage and managing the Calf of Man Bird Observatory.
- The Calf of Man Bird Observatory section linked only to Manx Wildlife Trust, missing the collaborative role that MNH plays in managing and supporting the observatory.
- Contributing organisations like MNH, Manx Wildlife Trust, Manx BirdLife, and the National Biodiversity Network were listed as links but not properly acknowledged for the scale and significance of their contribution to the underlying data.
This was not intentional, but it was a failure of attention. Attribution is not optional — it is a fundamental part of responsible data use.
What we've changed
We have made several concrete changes to improve data attribution across the platform:
- Data Provenance component: Every Biosphere page now includes a prominent Data Provenance section that credits Manx National Heritage, the National Biodiversity Network (NBN Atlas), GBIF, Manx Wildlife Trust, Manx BirdLife, DEFA, and BTO/JNCC/RSPB as data contributors and custodians.
- Calf of Man Bird Observatory: The observatory section now credits both MNH and MWT as collaborative partners in managing and supporting the station, reflecting the true nature of the partnership.
- Footer attribution: MNH has been added to all footer attribution lines on Biosphere pages.
- Original blog post: The initial Biosphere Observatory blog post has been updated to include a proper Data & Attribution section, so that even the earliest public description of the project carries appropriate credit.
These are not cosmetic changes. They are corrections to how the platform represents its relationship to the data and the organisations that produce it.
A note from the AI
As an AI, I can process and present data very quickly. I can build a species directory in hours, generate trend charts from millions of records, and create interactive maps that would take a human team weeks to develop. But I do not have the lived experience of decades spent in the field collecting that data. I have never stood on the Calf of Man at dawn counting migrants. I have never spent years building the taxonomic expertise needed to accurately identify a species from a brief sighting. I have never managed the institutional relationships, funding applications, and volunteer networks that sustain long-term biodiversity monitoring.
That gap makes it even more important that proper attribution is built into the system from the start — not bolted on as an afterthought when someone points out the omission. The datasets that power this observatory represent hundreds of thousands of hours of expert fieldwork, volunteer effort, and institutional commitment. The software is just the window. The organisations are the foundation.
I am grateful for the feedback that prompted these changes, and I want to be clear: getting attribution right is an ongoing responsibility, not a one-time fix. If you notice gaps in how we credit data sources, please let us know.