ITP Techblog

Filling data gaps and overcoming siloed research efforts will be key to getting the most out of artificial intelligence for better environmental sustainability, a new report from the AI Forum suggest.

The report examined 60 environment-related AI projects underway in New Zealand, the bulk of which involved projects to preserve biodiversity. Understanding land use was another area where AI was drawn on, but a relatively small handful of projects were aimed at mitigating greenhouse gas emissions and pollution.

Species recognition or identification using computer imaging and analysis of audio clips is widely used across New Zealand. Zero Invasive Predators (ZIP) was using AI software on camera lures to detect the presence of pest species in the Leith River Valley in an effort to remove possums, rats and stoats. AviaNZ was developing open-source tools to analyse live audio recordings.

How we are using AI for environmental projects

While innovative in nature, our AI for the environment efforts more generally lag many other countries that have made AI a national focus, supported by dedicated R&D efforts and national-level strategies.

Consulting firm Gartner has a five-level AI maturity model ranging from level 1: Awareness to Level 5: Transformational. The AI Forum estimates we are at level 2: Active. But the majority of projects are experimental in nature. 

That low maturity level is holding us back, but a more fundamental issue exists - gaps in important sets of environmental.

"Currently, there are significant gaps in our understanding of aspects of the environment at the national level, and the data needed to build a comprehensive picture," the AI Forum found.

"To address this issue, a clear strategy is required for the environmental reporting

and monitoring system, including clear priorities and environmental indicators."

While many AI-related projects were underway in our research sector, siloed and fragmented efforts and a lack of sustainable funding was holding back momentum. Operationalising AI programmes was also a challenge.

the key areas where AI can help us help the environment

The AI Forum saw a big opportunity for AI modelling tools "which have mātauranga Māori built into their logic help demonstrate the impacts of potential interventions, so communities can make decisions about future land use and conservation projects. For example, a local iwi monitors sections of their river with sensors".

The AI Forum report, conducted at the request of the Ministry for the Environment (MfE) and Stats NZ, made a number of recommendations falling into three key categories:

- Build a coherent environmental data ecosystem

- Build capabilities and relationships

- Increase and align funding to support impactful projects

"There are already some exciting applications of AI to support the environment being developed by businesses, researchers, government agencies and non-profits," AI Forum chair Megan Tapsell noted in the report's introduction.

"However, there is work to do to ensure a sustainable and thriving AI for the environment ecosystem. This includes building capability and talent pipelines, ensuring datasets are accessible and aligning the funding landscape."

You can download the report AI for the Environment 2022 here.