Although I’m still waiting for my own personal hoverboard (Back to the Future predicted we’d have them by 2015, I feel scammed), my news feed is now dominated by articles about new artificial intelligence (AI) technologies and what they are going to mean for our way of life.

Amidst the valid concerns and warnings about  the impact of AI on jobs, governance, copyright, ethics, and the very definition of humanity, there is also potential for these new digital tools to help us navigate the twin environmental challenges of climate change and biodiversity loss.

We have already seen how new digital technologies are making a difference in our transition to a nature positive future. They enable more effective data collection, automate time-consuming processes, enhance management decision-making, and support targeted monitoring for agreements and goals. This only reinforces the incredible potential of AI to better support biodiversity protection and buttress industry efforts towards meeting global conservation and sustainable development goals. We know “There is No Planet B” – but perhaps we’re improving our outlook for “Planet A(I)”.

“Found in Translation”: Why embracing multilingualism is critical for conservation

A worldwide study that investigated national reports on biodiversity conservation in 37 countries and territories where English is not an official language revealed that non-English-language literature is almost entirely neglected in global biodiversity assessments. This means that our existing global assessments overlook important conservation science by ignoring non-English-language literature. That is a lot of valuable conservation data and knowledge around the world that we’re not benefiting from.

Advances in neural machine translation, natural language processing (NLP), and other AI technologies to translate text from one language to another have reached a stage where they can now provide extremely accurate and fluid translations between languages. These translations can be easily scaled, and provide an efficient way to translate content into multiple languages that should give access to, and representation of, a wider global audience. We already have the capacity to train and customise AI translation models specifically for the conservation sector, but we need to think bigger. We should ensure the translation models work with specific audiences that have highly respected conservation knowledge, such as traditional owners and local communities. This would result in a more accurate, relevant and inclusive outcome.

We know that Indigenous peoples and Traditional Owners protect and steward over 80% of the world’s remaining biodiversity, and that Indigenous knowledge can help prevent environmental crises. However, much of the global conversation around land management and conservation excludes the skills and knowledge of Indigenous leaders and local experts based on the language they speak. With proper training, AI translation models could also give these experts access to participate in global conservation conversations in their own languages. Generative AI could be taught to employ models that better reflect Indigenous perspectives such as relational obligations and multigenerational responsibility when thinking about landscape management.

From Bites to Bytes: AI’s recipe for success in reducing food waste

Reducing food waste is essential in a world where more than 345 million people are facing food insecurity in 2023 (already more than double the number in 2020), and climate and conflict shocks have driven almost 900,000 to face catastrophic hunger.

In late 2022, AI purchasing systems to improve ordering accuracy were trialled in two large retailers by the Pacific Coast Food Waste Commitment (PCFWC), Afresh, and Shelf Engine. This resulted in a 14.8% reduction of food waste per store, saving 26,705 tons of CO₂e emissions from landfill. The savings in labour efficiencies from reduced ordering time and restocking, reduced shrink, etc, covered the cost of AI implementation and led to increased profits. The authors of that study estimate if the whole US grocery sector were to implement these solutions, 907,372 tons of food waste could be prevented, representing 13.3 million metric tons of avoided CO₂e emissions and more than $2 billion in financial benefits for the sector.

While this is a great start, the retail sector only contributes to only around 13% of the estimated billion tonnes of food wasted every year, while individual households contribute to over 60%. If households also had access to tools that reduce personal food waste we’d be on the right track. And that’s not out of reach – companies are working on smart fridges that scan your food, and you can already give ChatGPT a list of the must-use-today ingredients in your pantry to get some recipe ideas).

We also need to scale up agriculture-wide tech solutions to address waste on the other end of the supply chain. artificial intelligence can combine global data from remote sensors with local real-time data from farm equipment and soil health to design more effective regenerative agricultural practices, reduce food overproduction in high-income nations, and minimise premature harvest and food loss in lower-income countries. With an estimated 10% of global greenhouse emissions currently coming from food that is not consumed, this alone would make an enormous impact.

The Power of AI: A gamechanger for science, technology, communication, policy, economics…

Although ChatGPT can ‘hallucinate’ answers and writes well enough to fool scientists, there is also incredible potential for machine learning models to support research through verifying the statistical validity of data, or checking for patterns and inconsistencies across data sets. Computational modelling has long been used as a tool to study complex systems in biology, and the ability to build flexible frameworks that can access different types of data and find dependencies is exciting. Federated learning machines, for example, have recently been used to access small-scale studies to improve breast cancer treatment while patient data remained protected by hospital firewalls. Pollination itself recently announced investment in FLINTpro, a start-up with pioneering software that measures and manages carbon and natural capital across all land uses including forests, agriculture, grasslands, coastal areas and soils.

Another exciting development is the use of AI in satellite imagery, allowing us to detect changes in land use, vegetation, and forest cover. We can use it to detect and eliminate invasive species, and we can monitor and respond to the fallout of natural disasters. Clean Water AI uses a deep learning neural network to monitor water quality and detect dangerous bacteria and harmful particles in water in real time. The Tasmanian Birdsong Project is creating algorithms that are learning to recognise and count each bird and species in remotely captured audio recordings to help create a high resolution picture of what’s going on in local ecosystems. And the WildBook framework by non-profit WildMe uses neural networks and computer-vision algorithms to detect and count animals in images, which enables more precise estimates of wildlife population sizes.

We’re looking at a real potential for effective, automated and global biodiversity monitoring systems that deliver location and ecosystem-specific data in real-time, and then help to model the impact of the different management options under consideration. And while there’s plenty of anxiety in academia about ChatGPT being used to write student papers, on the flip side it is pretty amazing at summarising dense and complex documents in a way that makes them accessible to a wide audience. This could encourage people from different disciplines to work together on our conservation challenges, breaking down existing information silos and contributing interesting new perspectives and ideas.

Dipping into Digital Disruption: New tools and dynamic solutions

There are so many interesting digital developments for conservation already in progress! Blockchain is famed as a decentralised method of storing data that can improve transparency and traceability, and it can be used to help the most vulnerable of environmental practitioners. Philanthropists could  give money directly to local practitioners and reduce reporting burdens (currently, only 17% of global conservation funding intended for Indigenous Peoples and Local Communities actually reaches them), and Indigenous Peoples could track how their traditional knowledge is being used, as well as receive instant compensation as the knowledge holders.

Blockchain can also be used to create ‘smart contracts’ – programs that create trustworthy and self-executing contracts. For example, Cambridge Centre for Carbon Credits (4C) is using Tezos smart contracts with satellite and other remote-sensing data to automatically issue exchangeable carbon credits to landowners, and fund deforestation avoidance projects. The Lemonade Crypto Climate Coalition automates local parametric insurance claims, for example, automatically paying smallholder farmers after a pre-agreed number of days without rain. This allows them to immediately access tools to mitigate problems as they arise rather than having to file a claim after a disaster has occurred.

Cryptocurrencies allow us to experiment with new financial systems and mechanisms that incentivise nature-positive behaviour. For example, SEEDS (inspired by the Mayans’ use of cacao seeds as money) is a currency that incentivises collaborative and regenerative behaviours, and rewards activities that build community and regenerate our planet. Fishcoin incentivises supply chain stakeholders to share data from the point of harvest to the point of consumption – tokens move along the supply chain and those who make extra effort to capture and communicate data are rewarded, with the data then being used to improve decisions by seafood buyers and fishery managers. Crypto may also be more accessible to the ‘unbanked’ who are excluded from traditional markets due to distance from banking services, lack of traditional identification documents, high transaction costs, etc.

Non-Fungible Tokens (aka NFTs – essentially a unique digital asset) are opening up another alternative funding stream in the conservation space. Project Ark, for example, is a WWF initiative creating carbon neutral NFT collections of art and music which when purchased directly fund animal and environmental conservation efforts around the world. Moss is selling digital ownership rights to one-hectare portions of threatened lands in the Amazon forest, using the revenue to fund patrolling, satellite imagery and daily remote-sense deforestation monitoring of the project area.

Where to next?

We absolutely must be careful about the risks posed by AI. Among other things, we will have to manage issues like inherent biases in the programming, deepfakes, and bad data. We will have to work hard to ensure that the technology does not increase gaps in socioeconomic inequality. And we definitely want to avoid AI being used to drive more of our existing unsustainable business models and extractive consumerism.

But it’s also clear that we also have the potential to employ these new technologies to better protect our landscapes and reward local practitioners for their work. As we transform our gaze from pixels to planet, I’m optimistic we can harness new opportunities to help to build a nature-positive tomorrow.

About the author

Kirsty Galloway McLean is an Executive Director with Pollination Foundation in Melbourne, where she leads Ampliseed, a global learning and leadership network of conservation practitioners. She is a leader in environmental governance and knowledge management, including 15 years with the UN working on sustainable development and information sharing. Prior to joining Pollination, Kirsty set up the first globally distributed information exchange system under international law for the Convention on Biological Diversity in Montreal, led climate change and communications programs for the Traditional Knowledge Initiative at Japan’s United Nations University, and led an environmental and information management consulting company.