I also created a tool that finds species mentioned in the blog posts and creates a list at the end of the post that includes a link to a map of observations and a link to the species page on iNat.
I have a passing familiarity with how this kind of thing works, but no real expertise, the AI did all the coding. I liken it to knowing the car needs to be fixed but deferring to the mechanic because my fixes usually cause two new problems.
Looks good so far! Will the code update the field guide in real time as new species are observed on inat? Or will you need to run manual updates periodically to get new species added?
Several posts here are off topic and have been hidden. Future posts about AI / LLMs that represent general thoughts and opinions rather than how they specifically pertain to or have directly affected the content being discussed in a topic will continue to be flagged as off topic.
For general discussions about how AI / LLMs are used with iNat, the iNat API, or other third-party iNat tool development, anyone can feel empowered to create a new, general topic. For example, this one on how AI-assisted code could be a potential security issue when developing tools that request edit access to one’s iNat observations/IDs/etc.
Feedback on this or any other moderation decision can be raised with @forum_moderators in a direct message or over at Forum Feedback.
Edit: To improve clarity on what the topic is about, I also just updated the topic title from “A couple of AI builds that may be of interest” to “Park field guide examples using iNaturalist Data with AI assistance”.
What you did is called vibe coding. It is looked at more favourably than invoking LLM to summarise text.
“Vibe coded park field guide presenting iNaturalist data” would be a more representative title for your post.
Good term “Vibe Coding” The size choice was my attempt to give the uninitiated a way of sorting through a large number of species.The first sort is by the date of the observations. All the species are always available to view, but the default display is just the ones observed during the current month.
I tried using color as a key to sort bird species, but sexual dimorphism and the presence of more than one prominent color on most birds made this impractical for this use.
The main idea was building a tool that was easy to use and informative.