iNaturalist is excited to announce an award from Google.org Accelerator: Generative AI to help build tools to improve the identification experience for the iNaturalist community. The project we proposed builds off of our demo from last year to search iNaturalist photos with text. The award from Google.org provides $1.5M over 2.5 years plus access to Google staff to advise the iNaturalist team.
Our nonprofit mission is to connect people to nature through technology and advance science and conservation. We see this new opportunity with Google.org as a clear extension of the work that we’ve been doing for years to build better tools to connect people to nature. By using generative AI (GenAI), we hope to synthesize information about how to distinguish different species and accurately convey that to iNaturalist users. Instead of just offering AI species suggestions of what you saw, we want to offer a why as well. By providing explanations in addition to a list of suggestions, iNaturalist hopes to more effectively grow a skilled community of naturalists who have the information and tools to improve and vet the data on iNaturalist.
We explored this idea in this blog post last year using these frogs as an example. Currently, iNaturalist’s computer vision model can distinguish these frogs, but we don’t do a good job of explaining how. We’d like to use GenAI to tell people not just which frog it is but why it’s that frog.
Our goal is to build a working demo by the end of the year, and we’ll share more updates as the project evolves.
More generally, we're excited to use this grant to learn more about how AI is changing the technology landscape and how we can leverage these tools to enhance our mission and impact. As we learn about what AI tools are available to our nonprofit and how we might use them, we will continue to weigh ethical and environmental concerns.
We’ve tried to address some of the questions that have been raised on social media and in the forum in these FAQs.
FAQs
What is the history of iNaturalist and AI?
iNaturalist has been synthesizing photos and identifications using machine learning since 2017 to provide computer vision suggestions (a kind of AI) on the iNaturalist website and mobile apps. In 2023 we started incorporating observation location into our model training process, which resulted in the geomodel and range maps. All of the training of these models happens on two machines we own and control. Each time we update the model, we write a blog post about the new species that have been added thanks to the observations and identifications from the iNaturalist community. You can read the most recent blog post from April.
How is generative AI relevant to iNaturalist?
One of iNaturalist’s biggest strengths is its community of knowledgeable naturalists who spend time and energy helping identify observations. We know that this is a massive undertaking — especially as iNaturalist continues to grow, bring on new users, and support large-scale bioblitzes — and that without additional support, data quality becomes an issue.
Since its beginnings, iNaturalist has worked to leverage emerging technology (like computer vision, and even mobile apps in the early days) for biodiversity and conservation — and we think generative AI could be used to support the hard work done by folks making IDs on iNaturalist. For years we have grappled with how to surface the most useful identification tips shared by these members of the community, and we think that generative AI could provide a scalable way to synthesize and share useful information about how to identify the 100,000 different species included in our current modeling process. We are still exploring different approaches to this, but overall, the goal is to make observing and identifying on iNaturalist better and more enjoyable while also delivering more high-quality data needed for science and conservation.
How will you ensure that the identification tips are reliable?
We will incorporate a feedback process for the AI-generated identification tips so that we can maintain high standards of accuracy. Since this project is in its early stages, we don’t know exactly what this will look like, but we will share updates to be more transparent moving forward.
Ultimately, we want to continue synthesizing information across iNaturalist so that it’s more useful and accessible. We’ve been doing this with photos via our computer vision suggestions (and have a system to credit iNaturalist users whose photos and identifications were used in the model), and now, we’re excited to expand this tool to include text descriptions, too. We deeply appreciate your time and feedback, and we will continue doing our best to connect people with nature through technology.
Update on June 11, 2025 at 9:15pm EDT
We recognize that this announcement has caused a lot of upset and confusion among people who care deeply about iNaturalist. In the hundreds of comments, we hear that many people feel betrayed, disrespected, and without agency. We sincerely apologize.
With respect for the ethical and environmental concerns raised, we’d like to offer more clarity about our plans with this grant and program. We know that trust is not easily rebuilt, but we’d like to try.
First, to clarify some things we did not clearly communicate:
We are not replacing or changing the current human-curated system of identifications. Many of you have rightly pointed out that this is at the heart of iNat, and we wouldn’t dream of doing away with it. Anything we explore with AI would only be meant to enhance the experience of some users: by, for example, providing an interesting tip about identifying the species, suggesting that the photographer try to capture a particular aspect of the plant or animal that might aid a human identifier, or explaining the computer vision’s logic for the identification it’s suggested.
We are not giving Google special access to your iNat data, and we have no obligation to use Google’s infrastructure as part of this grant. Google is providing funding and advice on how to potentially leverage AI.
iNaturalist has successfully incorporated machine learning and computer vision (kinds of AI) since 2017, with a very small footprint (primarily three machines running in spaces we control). We understand that new technologies can have much larger environmental impacts, and we aim to quantify the environmental footprint of iNaturalist’s infrastructure for 2025 and beyond.
This grant funds our team exploring new ways to surface and organize helpful identification comments — and if the demo we create is not helpful, compromises data quality, has outsized environmental impacts, or is overall too flawed, we will not keep it.
This project is just getting off the ground, and we don’t know what it will actually look like to the user yet. We, along with many of you, are concerned that features like these could have a negative impact on iNaturalist, so our plan is to take an experimental approach to see if we can solve some of the core problems iNaturalist users have surfaced previously. If after testing and refining this experimental feature, we find that it reduces data quality and enjoyment of the platform — or its environmental impact is too large — then it’s not something we would continue with.
The iNaturalist community has always been a core partner in informing the direction of the platform, and we appreciate and hear your feedback (including but not limited to):
We understand that you are concerned about inaccurate, out-of-context and/or hallucinated information being provided by a tool like this. We agree that any AI-provided information be clearly labeled as such and that there be a way to upvote/downvote or otherwise flag potentially misleading information, and we are excited to explore more ways for how this may happen.
We hear that people would like to be able to understand and control how their data is used on and beyond iNaturalist. On iNaturalist, you own the copyright to your data, and iNaturalist Terms of Use prohibit the use of data for commercial AI training. Whether commercial AI companies are beholden to copyright law or terms of use or can argue “fair use” of publicly accessible data is being worked out in the courts and is something we are paying attention to.
Most importantly, the community will be involved in the creation of this and any major new features.
We will not be implementing any new changes to what you see on iNaturalist right now without involving the community. Our plan is to have an initial demo available for select user testing by the end of 2025. If you have more feedback, ideas, concerns, or questions you’d be willing to share, please use this form so we can more easily keep track of them.
Additionally, if you would like to attend a virtual Q&A session to have some of this discussion in real time, please let us know in this form (more details to be determined).
Again, the project funded by this grant is just getting started. We will proceed carefully and respectfully, and we’re immensely grateful for this incredible and engaged community.