From bin cameras to fish farms, AI can help save planet

We’re getting used to hearing that artificial intelligence might help solve many of humanity’s biggest problems, but it’s not immediately obvious that climate change is one of them.
AI is very energy-intensive — a search on ChatGPT uses roughly ten times more energy than one on Google — and this has led some observers to fear that it could actually impede the transition away from fossil fuels.
Michael Mann, a climate scientist, fears AI could be so energy-hungry that “even if we make substantial progress decarbonising our energy systems, countries like China will hedge their bets and say ‘yeah, we’re going to build renewables, but we’re going to continue building coal power stations too, because we need all the energy we can get’.”
Yet there’s also a more optimistic school of thought. Bill Gates, the Microsoft founder, has argued that the computer programs of the future will be so ingenious in helping businesses minimise waste that, overall, their impact on the planet will be dramatically positive, even if they use up lots of energy.
“Let’s not go overboard on this,” Gates told a London conference in June. “Data centres are, in the most extreme case, a 6 per cent addition [to energy demand], but probably [usually] only 2 per cent or 2.5. The question is, will AI accelerate a more than 6 per cent reduction? And the answer is: certainly.”
An analysis by PwC, the professional services company, predicted that the uptake of AI could reduce global greenhouse gas emissions by 4 per cent by 2030, while increasing GDP by 4.4 per cent.
It found, for example, that AI could help farmers plant their crops at the best times, help society make the most of renewable energy when it’s windy and sunny, and help reduce traffic congestion by enabling autonomous vehicles.
AI programs’ superpower is their ability to sift through vast amounts of data, then spot trends within it. In the third part of our series on the UK’s most promising AI companies, we look at firms that are already using this power to spot ways to cut waste, reduce carbon emissions and hasten the transition to net zero.
Founded in 2013 by Marc Zornes and Kevin Duffy, Winnow designs technology to help catering businesses cut food waste.
With a camera attached to the top of a bin, and scales attached to the bottom, Winnow’s AI program can identify discarded items of food with more than 80 per cent accuracy. It knows what a half-eaten taco looks like. It can tell the difference between mashed potatoes and chips. It looks at all these items and then tells kitchen managers what they should buy less of, saving them money and cutting food waste.
“Society has poorer-quality data on food waste than on any other environmental issue,” Zornes says. “Up until 2009 we didn’t even know how much food was thrown away in the world. When we can teach computers to see, it allows us to see more systematically things that we were unable to gather data on in the first place. That seems like a small thing but it’s actually a pretty powerful thing.”
Having deployed its technology in 2,700 kitchens in 85 countries, Winnow claims to save its customers $70 million a year.
Now that AI programs have been trained to recognise the contents of photos, other businesses are also using this ability to the good of the planet.
The co-operative Producers Direct, which represents more than one million smallholder farmers worldwide, has developed a program to analyse farmers’ photos of their crops. Shown photos from early in the Peruvian coffee-growing season, this program has been able to tell farmers how much coffee they are likely to harvest in coming months. That has helped them to hire the right number of labourers to harvest it.
Farmers feed this program data about their farms, and from that data it can tell them which farming practices are most effective. It can give them tips about when to plant their crops, what varieties are best suited to their climate and how much fertiliser to use.
AI is also peering into the lochs of Scotland, where Ace Aquatec is deploying cameras to analyse the health of farmed salmon.
The salmon industry has received much criticism over its impacts on the environment, including sea lice infestations inside fish nets, and algal blooms that result from the build-up of salmon faeces.
Ace Aquatec’s cameras can mitigate these problems by closely observing the fish and spotting signs of ill health long before any human would. They can work out when the fish are being overfed and they can spot sea lice long before they get out of hand.
“Just because of the sheer quantity of data we collect, it would be hard for our engineers to look through it all,” Keith Davidson, chief technology officer, says. “AI is a tool to work our way through all that data and look for patterns and indicators of ways to be more efficient, ways we might not have thought of.”
Computer vision is even being used to spot unexploded bombs under the North Sea, left there after the world wars. These bombs are an obstacle to energy companies seeking to meet the government’s objective of quadrupling the number of offshore wind turbines by 2030.
Until recently it took about two and a half years to identify these bombs and then remove them. Marine surveyors would have to spend months poring over seabed surveys, looking for indicators such as magnetic anomalies. Now, however, Rovco, a Bristol-based marine surveyor, has shortened that process to two-to-three months, using an artificial intelligence trained to detect bombs.
“We foresee a skills shortage in offshore renewables,” Joe Tidball, co-founder of Rovco, says. “So how do we take the number of people we have right now and allow them to do more work? It comes back to AI, allowing computers to look through multiple data sets at one time.”
Many environmentally minded investors are now asking themselves: where can I put my money to help save the planet rather than destroy it?
Not long ago wealth managers would have tried to answer that question by reading reports about companies’ environmental credentials, making a few phone calls to people in the know and then taking a punt on a company that seemed, on the basis of the scarce information available, to care about the planet.
Today, however, thanks to new regulations, companies are required to disclose a vast amount of information about their impact on the environment — far more than wealth managers can easily or efficiently comprehend.
That’s where Arabesque Asset Management comes in. Founded in 2013 by Omar Selim, a former Barclays managing director, the company has developed an AI program that can quickly assess the wealth of data that companies are required to disclose. It can then help wealth managers to pick stocks that are both environmentally sustainable and profitable.
“Behind closed doors, portfolio managers will say that sustainability is a pain,” Selim says. “They will say, ‘I want to invest the money as [well] as I can, but now I have to integrate score cards and carbon and human rights and labour rights, and I’m not an expert on any of those topics.’
“But it doesn’t have to be a pain, because machine learning can make it much more efficient.”
Selim claims that funds managed by Arabesque AI outperform those managed by humans 70 per cent of the time. “What you used to do as a portfolio manager — reading the papers, chatting to a few people — that is gone. The new way is machine learning. With machine learning, you can process billions of data points.”