AI, iNaturalist and Education

Good Morning All

What
we are a company that makes environmental education products, specifically water quality monitors. We are funded by the US Department of Education and just released a companion platform that helps educators create STEM field or classroom projects. This tool now uses AI and I am happily wading into the controversy of using AI in education AND using AI with iNaturalist.

I am not promoting my product here, I am explaining how we work with iNaturalist and AI and try to do it in a safe and constructive way.

I have been very hesitant to use AI in our development, outright critical, actually. I think AI has the potential to cause a lot of harm in education and the risk is not discussed a lot or is not heard over the hype. So, for me to go into a development project with AI was not a straight road, we thought about this a lot.

What does this have to do with iNaturalist?
Our new system allows educators to build STEM programs (“Labs”) that use specific natural habitats, e.g. a lake, creek, pond that the students know and identify with. The selection is done using a map and we then use iNaturalist’s API to pull in animals, plants or fungi that have been observed there. We then prompt an AI to integrate class grade, STEM topics, reading level, NGSS, etc and generate a text block or a question set based on both the identified organism and the teacher’s directions. The outcome is STEM content that is acceptable in schools/teaching but not from a generic textbook but based on very specific local environments that the students know.

Why?
Letting AI loose on students isn’t my plan at all. We will always have the educator in the loop and help them build amazing content. That rule is set in stone. And we use iNaturalist because we want to draw on the amazing community, promote citizen science, foster engagement and create a starting point for the educator that is not hallucinated.

What we don’t do:
the tool is a one-way-street. We don’t feed information back to iNaturalist, we don’t build our own AI or models on it, either. We will be looking into submitting data eventually but only after very careful reflection and hopefully input from the iNaturalist community.

Now what
Well, I started the company to help teachers, students and citizen scientists in the only way I know how, build better tools. I’d love feedback from you all how I can do better. If you have ideas, concerns, objections, praise (yes!), please let me / us know.

If you want to know more
If you really want to know or you are a STEM teacher who wants to try our systems out (for free, btw), let me know, I can give you a demo.

See you outside.

Dietrich

“In the end we will conserve only what we love, we will love only what we understand, and we will understand only what we are taught.” Baba Dioum

2 Likes

As an educator who works in wildlife and conservation, my main goals are to teach my kids about the environment and to hopefully inspire them to live a life with sustainability and conservation in mind and in action, if not directly going into the field.

This is my job. Why do I need an AI to do this for me? Does your model take information from preexisting sources? If so, this is pointless. In this field, you are learning constantly about the species around you. You are retaining this information and presenting it to your audience. If I don’t know what species a bird I see is, I’ve already got all the tools I need to figure this out. Regarding the creation of questions and quiz material, having AI create this more me feels like it would be way more work when I already know what I want to ask and can just type it out myself.

As you have mentioned, this is a hot button topic. I am beyond skeptical at the thought of integrating anything AI in my lessons. I’m interested in why this would be beneficial over the way things are now. I’m also interested in your company’s sustainability practices.

9 Likes

apologies for the delay and the unexplained unlisting, please give me a few minutes. I logged on only a moment ago.

Thank you for your thoughtful response, this is the kind of feedback I was hoping for. Let me answer firstly that not every tool is for every person and as an expert educator in wildlife and conservation, our tool may not be useful to you.

“[…] my main goals are to teach my kids about the environment and to hopefully inspire them to live a life with sustainability and conservation in mind and in action, if not directly going into the field.”

That is exactly what our goals are and how do can we empower STEM educators who may not be experts in wildlife and conservation to instill the same passion into their students? We work with educators from Grade 5 onward who have to adhere to NGSS (or equivalent), work solo with 30+ students and achieve challenging education goals. They are amazing educators but not always experienced wildlife experts and our tools aim to empower them to take the students outside and contextualize what the students see, touch and experience with the STEM content they are asked to teach.

Why do I need an AI to do this for me? Does your model take information from preexisting sources? If so, this is pointless. In this field, you are learning constantly about the species around you. You are retaining this information and presenting it to your audience.

These are valid points, why do we need “AI” to do this, and I struggled with this for a while. In fact, we are “late to the party” because we took a few big steps back in releasing AI. We are now confident that we use AI in a constructive, helpful way that enables the teacher, not reduces their role.

If I don’t know what species a bird I see is, I’ve already got all the tools I need to figure this out.

When I was an undergraduate student (Biology), I went on a field trip to a local wetland habitat. The professor knew every single plant, animal and fungus and how they interacted. It took him decades to study and research. He had all the tools he needed in libraries full of textbooks, he would have not needed iNaturalist, it wouldn’t be the right tool for him, similar possibly that our new tool isn’t the right one for you.

Regarding the creation of questions and quiz material, having AI create this more me feels like it would be way more work when I already know what I want to ask and can just type it out myself.

Thats a real risk we are working with and we have the tool currently in use with 5th grade STEM teachers who are not wildlife experts to really understand where we add value and where we distract from their core mission. Using AI is incredibly risky and our job is to provide tools that empower rather that frustrate or confuse.

As you have mentioned, this is a hot button topic. I am beyond skeptical at the thought of integrating anything AI in my lessons.

Preach it, me too. See above, I think we found a compromise or I wouldn’t have released it.

I’m interested in why this would be beneficial over the way things are now. I’m also interested in your company’s sustainability practices.

So, nothing beats a demo and of course I invite you for a zoom call or something to show you. But I want to make sure that I keep this thread as “non-promoting” so, I am careful around links etc. If you genuinely want to know more and have a chat (public if you want), let me know, I’d love that.

Dietrich

2 Likes

I needed to take a little time to investigate, and my conclusion is that you (a) are not promoting commercial content and (b) are promoting a conversation about a specific nature-related content, and (c) are disclosing your personal connection, all of which is appropriate. I initially hid the thread as a precaution but I’ll be un-hiding it in a moment.

I do think your initial post would benefit from at least one real practical example. It might be a me problem but I find the exercise and its purpose difficult to follow.

2 Likes

Thank you, I appreciate the response. I am putting a short video demonstration together this weekend, would it be acceptable to link it? Again, I don’t want to (ab)use this forum to promote.

Thank you for your thoughtful response! This did put it in a way that makes the purpose more clear to me. I appreciate that you seem to be considering all the angles and risks from a non-partisan standpoint!

1 Like

I’ll start with a caveat: I’m not a classroom teacher. I spent 25 years in education, but it was outside the classroom. However, that gives me some perspective that might be useful, especially because my job required me to do a fair amount of curriculum development.

As I understand your product, you are using generative AI to generate a customized curriculum guide to be used by overloaded and overworked classroom teachers. In short, while you say that you’re building “tools,” from my perspective what you’re actually doing is curriculum development – your product is designed to support and supplement existing curriculum.

I’ll get to generative AI after a digression about curriculum. Bear with me.

Curriculum, and curriculum supplements, are always built on assumptions, and always contain biases. There is no such thing as a value-neutral curriculum (or curriculum supplement). I remember in my graduate coursework in curriculum development having to identify the educational assumptions and biases in curriculum – it was eye-opening. The best curriculum materials are explicit about their assumptions, and try to acknowledge their biases. You want to be upfront about it. So, for example, as soon as you say “STEM,” that usually comes from an essentialist educational philosophy. But I’m also sensing elements of an educational philosophy that aims to engage students in changing the world around them. So maybe something of a mixed educational philosophy here.

When I do curriculum development, I’m always looking at how the constituent materials support or conflict with the guiding assumptions. So for example, when I was part of a team developing an ecojustice curriculum for grades 6-8, the goal was to help students come to a better sense of themselves, and to provide support in and motivation to create positive change – a mixed educational philosophy. We made deliberate choices in what technology we used – hands-on building projects, citizen-science projects, direct engagement with a wider community, etc. – and to fit our educational goals and assumptions we were deliberate about our use of technology, so e.g. we didn’t show video explanations of topics (tends to make students passive rather than active) but we did bring in guest speakers via video conference. Every choice was deliberate (to the best of our ability).

Now we get to generative AI. You express some ambivalence about using generative AI in your product. I guess I would want to know how using generative AI accords with your educational philosophy and your other curriculum assumptions. Maybe you’ve already done so, but it seems like it might be helpful to work with a specialist in curriculum development, to help articulate the educational goals and assumptions of your product.

This is all part of a wider societal reflection about how generative AI is already affecting all kinds of learning situations. Students are already using generative AI in many contexts. Teachers and other educators, who are typically underfunded, underpaid, and overworked, are already using generative AI to help them get through the day in their underfunded jobs. Not much point in closing the barn door once the horse is already gone. So now we’re all starting to reflect on what assumptions generative AI injects into the educational process, and how the use of generative AI is influencing educational goals. Because while I personally don’t want to use AI-generated curriculum support materials in my educational praxis, the harsh reality is that educators have less and less time to do more and more, and it might be that I don’t have much choice…but I do have choice about how and why I use AI generated materials, and how it affects my students. And it would be really helpful to working educators if the tech people providing AI generated educational products were more aware of these issues.

5 Likes

Thank you, this is a thoughtful comment.

I think your main point is fair. Even if we call Navigator a tool, it still shapes how material is framed, so it does carry assumptions and isnt neutral. We should probably be clearer about that.

At the same time, our goal is not to replace teacher judgement. It is to help teachers build local, place-based STEM content more quickly, especially when they are not specialists in local ecology. The teacher still decides the place, the goals, the level, and what finally gets used. Teacher control stays in the foreground.

We were also pretty hesitant about generative AI for exactly these reasons. We are trying to use it narrowly, as a drafting aid around real observations and educator intent, not as an authority and not as a substitute for field experience or teacher expertise.

And of course the practical side matters too. Teachers are overloaded, and that is the environment they are working in every day. That doesnt make every use of AI good, but it does mean useful tools have to be judged in that real-world context.

So thank you, really. This is usefull feedback, and I appreciate you taking the time to spell it out. This is the feedback I was hoping to receive!

And of course, if you are curious how we are doing this in practice, please let me know and I am happy to show it to you.

Dietrich

PS.
Wrt “philosophy”. I started my company with the goal to help re-define our relationship to the environment, making it personal and deliberate. If we understand the environment that touches us, it automatically becomes part of us and we are more committed to take care of it. At least thats the plan ;)