A Deeper Understanding of Wildlife and Nature

Each species featured on MUUUH Nature Documentaries is studied through a structured process that combines field observation, AI-assisted literature research, and validation through direct observation in nature.

The more I integrate AI into my workflow, the more important it becomes to verify information through my own knowledge and observations.

Key Takeaways

Field observations trigger the study of a species.
AI is consulted through a structured natural-history questionnaire.
The generated answers become research notes used during field observations.
Observations verify, refine, or challenge the information.
Only validated insights are incorporated into species profile pages.

Research Method

The steps below show how field observation, AI-assisted literature research, and field validation contribute to a deeper understanding of the animals and plants I encounter.

Field observation – I encounter a species I am not yet familiar with and document it with photographs, audio recordings, or field notes.
Questionnaire – A structured set of natural-history questions is used to consult AI about the species’ ecology, behaviour, and life history.
Research notes – The generated answers are saved as markdown notes and used as reading material while observing the species in the field.
Field validation – During later encounters, I compare my observations with the information in the notes and refine my understanding of the species.
Species pages – Verified insights and field observations are incorporated into the species pages published on this website.

AI Questionnaire

To better understand each species, I consult AI through a carefully designed questionnaire instead of asking ad-hoc questions. This approach ensures that key ecological aspects are explored systematically for every species.

Question Framework

The questionnaire includes roughly 20 questions covering key natural-history themes, including distribution, ecology, adaptations, behaviour, and reproduction.

Example questions include:

What habitat types does the species typically occupy?
What predators commonly take the species at different life stages?
What traits help distinguish the species from similar species in the field?
When does the main breeding season occur across its range?
What ecological role does the species play in its ecosystem?

Guided AI Responses

To keep the responses focused and useful, the questions are sent together with a carefully designed system prompt.

The prompt instructs the model to:

answer as a natural history assistant
prioritize ecology and behaviour in wild populations
keep answers concise and structured
avoid advice related to gardening, pet care, or captive husbandry
prioritize information relevant to Québec and northeastern North America

Example Q&A

The following example illustrates how the questionnaire generates structured explanations for a species, using the House Finch (Haemorhous mexicanus) as an example.

Question

What are the main differences in appearance between males and females of the House Finch (Haemorhous mexicanus), if any?

AI Answer

The House Finch (Haemorhous mexicanus) exhibits sexual dimorphism, meaning males and females have distinct differences in appearance.

Male House Finches: Males are notable for their bright coloration, typically displaying vibrant red plumage on their heads, throats, and chests. The intensity of the red can vary, ranging from orange to yellow, depending on their diet during molting. The rest of their body is mostly brown with streaks.
Female House Finches: Females, in contrast, lack the bright red coloration. They are generally brown and streaked all over, with a more uniform appearance. Their plumage is more subdued, with streaks on the breast and belly, which helps them blend into their environment.

These differences in coloration are primarily due to the males' role in attracting mates, as the bright red plumage is a signal of health and vitality. Females, being more camouflaged, benefit from reduced visibility to predators, especially during nesting.

Validation

The overall description is accurate, but the red coloration of males is typically most pronounced on the forehead and upper breast. The body plumage also appears grey-brown rather than brown in many individuals, a distinction that becomes evident during close observation.

In my own observations in Québec, males consistently show red coloration, and I have not encountered individuals with orange or yellow tones. This may reflect regional variation in diet or the populations present in the areas where I observe the species.

Research Notes

These notes are not intended for direct publication. Instead, they serve as a structured reference that I consult while identifying and observing species in the field.

How the notes are used:

Reading – In the field, I review research notes relevant to the species I am observing.
Observation guidance – The notes highlight behaviours, identification traits, and ecological interactions to watch for.
Verification – I compare field observations with the notes, confirming, questioning, or refining the information.
Content extraction – Insights from the notes that are confirmed by my field observations may be incorporated into published content.

Over time, this process transforms the initial AI-generated answers into a curated body of natural history knowledge shaped and verified through direct field observation.

AI Limitations

In this workflow, AI serves as a literature research assistant rather than a source of verified knowledge. The generated answers provide structured summaries of commonly available ecological information and serve as starting points for further investigation.

To evaluate this approach, I compared the generated research notes with real-world knowledge in two ways. For species I already know well, I verified the AI responses against my own knowledge. I also shared notes with experienced researchers for feedback and verification.

These tests showed that AI often produces generalized ecological descriptions. When reliable information for a question is limited, the model may speculate or generate plausible but unverified explanations.

For this reason, the questionnaire and system prompt have gone through multiple rounds of refinement to improve the quality and relevance of the generated responses. Even with these improvements, the answers still require verification through direct field observation.