We’re often told that artificial intelligence poses an existential threat to humanity (at worst) or is poised to steal our jobs (at best). But what if the technology can provide solutions to some of the world’s biggest challenges — such as combatting the climate crisis or helping those in underserved communities?
Forget the doom and gloom—here TI reports on three different AI firms using AI for good: AiDash for biodiversity and property development, AutoDesk for coral reef restoration, and Accion, which is using it to help marginalised communities and even ensure that the tech itself isn’t discriminatory.
According to figures from the Biodiversity Intactness Index (BII), the UK is one of the most biodiversity-depleted countries in the world. However, a BII study found that the UK has just half of its biodiversity left, leaving it languishing in the bottom 10% of the world’s countries.
AiDash is a firm focused on making infrastructure industries climate-resilient through satellites and AI. For biodiversity, it uses satellite imagery to get a bird’s eye view of what’s happening on the ground.
Working with satellite data providers, once an area is requested for analysis, AiDash will simply source an image that might already have been used for analysis.
According to Shashin Mishra, VP of EMEA at AiDash, the firm’s model seeks to identify exactly how many habitats there are by labelling each one. It then detects any invasive plants that pose a risk to native species, offering insights into the health profile of fauna.
This not only helps those fighting to improve the country’s biodiversity but also helps industries such as the UK’s property sector, where they must legally meet Biodiversity Net Gain (BNG) requirements before building on a site.
BNG requirements mean developers must ensure their sites result in more or a better quality natural habitat than what was there before.
“When a property developer starts a project, they’re going to have a baseline preliminary survey done on the site,” explains Mishra.
This is typically done by an ecologist, but Mishra argues that these kinds of surveys lack accuracy and detail.
“Typically, what these surveys do not tell us is how much biodiversity is there and what kind of condition it might be in,” he says. “That’s what we deliver. They can see that as a report, and they can open it up on the web and interact with the system and see exactly what’s there.”
If we’re looking for true biodiversity, there are few places more diverse than coral reefs. Based on species diversity, for example, the Great Barrier Reef – the world’s largest coral reef – is one of the most diverse habitats on the planet.
Coral reefs also play a key role in our food security, with over one billion people reliant on reefs to support their food supply chain in some form. A quarter of marine species live in reefs. Yet, according to the UN, between 70% to 90% of coral reefs could be lost within the next 26 years.
“Reefs are critical carbon sinks,” says Nic Carey, senior researcher at Autodesk Robotics, “and they contribute to maintaining ocean pH levels and sustaining marine health, even well outside their immediate environment.”
Autodesk Robotics researches the future of robotics in industries such as manufacturing, architecture, construction and nonprofits with an underlying ethos of creating a better world.
To save coral reefs, Carey says the world needs to restore and plant hundreds of hectares per year, minimum. Currently, the world’s combined efforts fall far short, with only about one hectare planted per year.
This “serious mismatch in necessary goals comes from the fact that “there simply isn’t enough human workforce to address the scale of coral seeding and planting,” she explains
Reef monitoring is already abundant, and AI plays a key role in helping scientists predict bleaching patterns and the depletion of biodiversity. But, according to Carey, coral restoration projects are still underserved by automation.
Most conservation-restoration programmes are still done completely manually through coral seeding and planting in nurseries. Although seeding and planting is not a difficult task, Carey says, it i slow and extremely labour-intensive.
In 2020, Autodesk Technology Centre in San Francisco partnered with Australian restoration firm Coral Maker in order to test AI-powered robotics to automate the process of seeding and propagation of corals
By using off-the-shelf robots and sensors alongside open-source or commercially available software, the partnership is able to automate coral planting and seeding.
“Robots offer a significant force multiplier, but they’re also ideally suited to automation because their tasks are repetitive,” says Carey
With a combination of robotic arms and image sensors, the robot cuts fragments of an existing live coral into smaller pieces, glues them into plugs and implants them into limestone ‘skeletons’ to eventually grow into a bigger colony.
The skeletons are in the process of being mass-manufactured using recycled stone composite. With this and the robots, AutoDesk says it has the potential to scale 10,000 skeletons per day, each with the capacity to hold six to eight ‘fragments’ and ultimately help bring biodiversity back to the oceans.
Washington-based international nonprofit Accion uses technology such as AI to support small business owners, smallholder farmers, and women who are most impacted by climate change, economic instability, and conflict with financial services.
For instance, Apollo Agriculture, a portfolio company of Accion Venture Lab, uses technologies to support small-scale farmers across Africa in increasing their profits and farming more sustainably.
Its AI and automated operations enable farmers to access optimised financing, farm products, digital advice, and risk management solutions. For example, its predictive repayment propensity models help Apollo access risk in the market where farmers tend to be excluded from formal financial systems and absent a formal credit history.
Accion also includes the think tank Center for Financial Inclusion (CFI). It has published a toolkit for venture capitalists to vet start-ups and ensure that they do not use data that can lead to discrimination against people who are already economically vulnerable.
“The topics we’ve been examining are really with the aim to help build trust and digital finance by bringing more transparency and accountability for data practises that providers are using,” explains Alexandra Rizzi, senior research director, consumer data opportunities and risks at CFI.
“That includes issues around privacy, around AI, around data governance, and that transparency and accountability could be brought by different actors, whether those are impact investors who are asking before they invest in a particular fintech: How are you building your models? Where are you getting your data from?”
During the Covid pandemic, Rizzi recalls when many businesses had to close and needed access to relief because their income had essentially come to a halt.
“A number of countries did cash transfer programmes, giving out payments, but in many of those emerging markets, there were not strong social registries of who should actually be targeted,” says Rizzi.
The West African country of Togo launched a programme called Novice, which used an AI model that crunched satellite data, as well as mobile money transactional behaviour, used as a proxy for someone’s economic status and looked at the poverty levels of certain geographies.
“In the absence of a robust social registry database to say, well, this individual needs this economic payout, they were able to target individuals and build an AI model that leveraged the data.”
“So there are all kinds of incredible use cases I think that AI is already achieving,” says Rizzi. “But we also want to make sure that it doesn’t lead to unintentional harm or further exclusion, and so our work has been to raise very clearly the opportunities that are out there, but also make sure that these fintechs are aware of some of the potential risks.”
For instance, the CFI knows that in certain geographies there are gaps between men and women and ownership of phones, and so they have very different data profiles.
“So, our AI models need to adequately acknowledge these kinds of gaps in data trails between men and women and also with other excluded marginalised groups.”
Subscribe to our Editor's weekly newsletter
Share
