🌿 WildSphere AI: Turning Open-Weight AI Into a Reason to Explore the Real World

WildSphere AI, a project for the Hacktoberfest 2026 DEV Challenge, aims to blend AI with the real world by transforming open-weight AI technology into an engaging tool for exploring nature. In today's digital age, people often overlook the fascinating world around them, missing out on the beauty of birds calling from trees, butterflies fluttering in gardens, or previously unnoticed plants.

WildSphere AI seeks to bridge this gap by offering a nature field guide that identifies, explores, and educates users about the living world, encouraging them to spend more time experiencing it offline.

The AI-powered nature guide utilizes Gemma 3 4B Instruct, an open-weight model, to identify plants, insects, and animals from photographs. Instead of providing a simple name, WildSphere AI offers AI-guided suggestions for users to investigate, acknowledging that two species can look similar and that one photograph may not be sufficient for a confident identification.

By treating AI suggestions as leads to investigate, rather than unquestionable facts, WildSphere AI promotes responsible AI use, being honest about uncertainty, and fostering a deeper understanding of nature.

One of the most exciting features of WildSphere AI is the Interactive World Biodiversity Atlas. Users can select a country on an interactive world map, which automatically loads biodiversity observation records for that country. Users can then explore these records and filter them by category, such as animals, birds, and plants.

Users can follow their curiosity into more detailed information by connecting to individual species pages. The atlas uses observation records, including data from iNaturalist where available, and acknowledges that a record only shows that a species has been observed or documented, not that it is native, common, or present throughout the country.

The Species Discovery Pages provide additional information about identified species, such as the species name, scientific name, observation counts, and extra details when available. WildSphere AI distinguishes between available and unavailable information, ensuring that users receive accurate and complete data. This feature connects the image guide and country atlas to a more comprehensive discovery experience, enabling users to identify a possible species, explore its records, and learn more about it.

The Bird Sound Explorer allows users to upload audio recordings and explore bird-call identification. The interface supports common audio formats and offers country and habitat context. The intended classification system is BirdNET, which is designed for bird-sound identification. However, WildSphere AI acknowledges that an ordinary language model or audio-file upload alone cannot reliably identify a bird call.

When classification is available, its output should be treated as a prediction that may need verification, encouraging users to verify the results.

Lastly, WildQuest is designed to transform ordinary outings into manageable nature missions. Users can request a guided outdoor activity by describing the type of outing they have in mind, optionally adding a country or something they'd like to notice. Missions might encourage users to listen for different bird calls, observe leaf shapes, notice pollinators, or pay attention to environmental changes.

The goal is not to turn nature into another checklist but to give people a gentle starting point when they are unsure what to look for. The missions should also respect wildlife and the environment, ensuring that users do not disturb nests, feed wild animals, touch unfamiliar plants, or approach dangerous animals.