You know me. And for those who do not know me yet – I love to share my stories.
Subscribe
Icon Rounded Closed - BRIX Templates

Sustainability and AI: How a new wave of projects are using AI to weed out waste

Sustainability and AI have been dominating the news in recent years. Much has been written about the higher energy consumption of AI and the associated risks to the planet. But could AI also help to drive efficiencies to make businesses more sustainable? Projects in the food sector provide a fascinating insight into what lies ahead.

Email Icon - German Ramirez

Subscribe to my newsletter

Lipiscing elit. Mi, diam venenatis amit.

Thanks for joining our newsletter.
Oops! Something went wrong.
Sustainability and AI: How a new wave of projects are using AI to weed out waste

Sustainability and AI have been two of the biggest corporate hype cycles of the last decade. Tesla rode a wave of sustainability hype to become a major player, while the AI boom, especially after ChatGPT’s launch, propelled NVIDIA to a market cap of over $3.24 trillion. Given the momentum in both fields, it’s no surprise that companies are exploring how AI can enhance their sustainability strategies. But do these initiatives amount to anything more than just good PR? Let’s take a look at some of the most interesting projects at the intersection of sustainability and AI in the food sector. 

AI in Agriculture: Cutting costs and carbon

Firstly, let’s be clear — AI poses legitimate environmental concerns, with queries up to 10 times more energy-hungry than standard web searches. Companies must carefully weigh the environmental pros and cons of adopting AI for sustainability and be transparent to avoid greenwashing.

These concerns notwithstanding, some brands are using AI in innovative ways to boost the sustainability of the food industry. The meals on our plates have a higher environmental cost than we might think. Data shows that food production, including crops, livestock, fishing, land use, and the food supply chain, accounts for 52.3 billion tonnes of CO2 per year, which equates to 26% of all carbon emissions globally. For comparison, the emissions of the entire aviation industry — known as a highly carbon-intensive sector — currently amount to 2.5% of global emissions.

In terms of crop production, one of the major contributors to the footprint of food is the widespread use of herbicides and pesticides, whose manufacture, transportation and disposal all lead to emissions. Anyone who has gardened knows that weeds can be a real challenge, competing with plants and crops for space and nutrients. In an industry with tight margins, many farmers are forced to spray their crops indiscriminately with herbicides. While it is estimated that only 3% of cultivated farmland contains weeds, 100% of it typically needs to be sprayed to contain them. This not only has big environmental costs, it also poses a large financial strain on farmers who waste millions of dollars spraying herbicides on land without weeds each year. But unless they are prepared to spray each weed individually, what is the alternative?

Weeding Out Waste: Precision AI’s smart sprayers

The Canadian company Precision AI is using IoT, computer vision and AI to tackle the problem. It has developed a specialized unmanned aerial vehicle (UAV) that can carry over 375 liters of herbicides for up to five hours and spray it in a selective and targeted manner at patches of weeds. Visually, the device looks a bit like an oversized paraglider, combining a large, oblong parachute with a small and efficient 100 hp motor underneath.   

Source: Precision AI

It uses a dataset of photos of similar crops to train a model which can differentiate weeds from plants in real-time. While flying, the onboard system categorizes between various types of vegetation and targets the spray over areas where weeds are detected, reducing the financial and environmental costs of applying herbicides. The company claims that it costs just $7.04 to spray a hectare of land with its latest vehicle, compared to $15 - $32 with other drone solutions.

When attempting to deploy IoT devices in agricultural settings, tech companies often come up against a snag: no internet access. The same challenge is true here: UAVs may need to operate in remote areas where signal is weak or non-existent. In response, Precision AI has implemented an edge computing model, whereby almost all data is processed on-device. This removes the need for internet access and has the additional benefit of making weed identification much faster, as data does not need to be passed to an external data center for processing.

Source: Precision AI

Waste Not, Want Not: AI tackles food waste

What about the other end of the food supply chain — consumption? Eurostat estimates that about 10% of the food made available to EU consumers is wasted, with over 58 million tonnes generated annually. There are two sides to this type of food waste: commercial catering/hospitality, and the food in our homes. 

When it comes to commercial catering, a key challenge in minimizing food waste is to accurately estimate demand. The airline industry is a case in point. At a time when flights are far cheaper than in the past, 3-5% of customers book flights and fail to show up on time at the airport. This makes it harder to gauge how much food will be needed, with approximately 20% of all in-flight food wasted each year.

Source: KLM Newsroom

In response, KLM teamed up with Dutch AI consortium Kickstart AI to use machine learning and predictive AI to get more accurate estimates about passenger numbers. The model uses historical data from the 17 days prior to departure until 20 minutes before takeoff to estimate meal demand. The airline says that the implementation of the system has resulted in 63% less food being wasted than if it had provided catering for every booked passenger. Kickstart AI has now applied similar strategies to reduce food waste in other settings, such as at food banks and for the supermarket company Delhaize.   

The fight against “fridge blindness” 

Other companies have explored ways of using AI to encourage consumers to waste less of what they buy. The condiment maker Hellmann's, famous for its mayonnaise, has developed a new web app called “Meal Reveal” in collaboration with Google. It enables you to take a photo or video of the current contents of your fridge. The app then uses computer vision technology and generative AI to analyze its contents and suggest possible recipes for meals using the ingredients. Hellmann's claims that the app helps busy professionals deal with a phenomenon it calls “fridge blindness”, whereby people come home tired after work, look in the fridge, and can not think of something sensible and nutritious to cook.

Source: Hellmann's Meal Reveal

Given that it is a web app, no installation is required and it can be used on any smartphone. When I pointed it at my fridge containing a random assortment of leftovers and took a quick video, within about 15 seconds it asked me to confirm the ingredient list and suggested that I cook smashed potato nachos. Overall, I was quite impressed with the performance of the app. The recipe was not amazingly creative, but it was a decent suggestion served up impressively fast. The look and feel of the app is polished and the UI is user-friendly too. 

Despite these benefits, I have some reservations. Some ingredients, like plums, were falsely identified and had to be manually removed. The suggested recipe used only a small portion of the produce, and the app couldn’t determine which ingredients needed to be used up first. A future iteration could integrate with supermarket loyalty cards to access dates of purchase, but that would require a longer setup process and customer account.

More fundamentally, is the computational and energy cost of AI video analysis justified? How much CO2 is emitted compared to the food savings? This is a difficult question that can only be answered with quality research. Hellmann’s could address this by being transparent about the carbon footprint of its app and comparing it to the average environmental cost of food waste.

AI’s greatest impact may be hidden from view

Having taken a closer look at these projects, it is hard to avoid the conclusion that some of the biggest gains will be made in the supply chain rather than at consumer level. For me, it comes down to the ratio of environmental impact to computational load. On a farm, using AI could result in thousands of hectares being farmed more efficiently, fewer chemicals entering the groundwater, and a lower carbon footprint. A catering company, restaurant chain, or airline could avoid wasting thousands of meals each day. In comparison, as fun and well executed as the Hellmann's app is, the cost-benefit calculus seems a lot less clear cut. 

This insight has implications for brand building. Some of the biggest environmental benefits of AI will not be directly experienced by consumers. As a result, it will be up to the communications teams at companies to draw attention to these initiatives in creative and compelling ways and use them to build brand value. In an economy where over half of GenZ and millennial employees say that they actively research a company’s environmental impact and policies before taking a job, sustainability is also becoming an essential part of employer branding. 

Green Advantage Keynote

Businesses that act swiftly and decisively on climate change can secure a long-term competitive edge. Join me for the Green Advantage keynote and inspire your audience to embrace a sustainable transition. For further details and contact info, click here.

Image: Photo by Google DeepMind

Thanks for joining our newsletter.
Oops! Something went wrong.