Machine learning bots at Ocado. Handout.
Slaves to the algorithm: robots at an Ocado distribution centre

Ocado customers will be pleased to know that the online supermarket’s warehouses generate virtually no food waste.

This is because Ocado uses machine learning to anticipate demand and to decide exactly how many products it needs to stock. A similar logic applies to its other processes: at its distribution centres, cohorts of robots criss-cross huge metal grids in an algorithmic dance that enables a 50-item order to be sorted within five minutes. As with the food, not a movement is wasted.

Such technological prowess helps the bottom line, of course, but it also reduces the company’s environmental impact. And Ocado is not alone. As consumers become more selective about shopping sustainably, businesses are trying harder to eliminate waste from their processes. That in turn is leading to increased take-up of artificial intelligence: according to tech market advisory firm ABI Research, the number of AI-enabled devices in industrial manufacturing will reach 15.4m in 2024.

From a circular-economy perspective, this is good news. If the unsustainable linear economy is about take-make-waste, technology that minimises or even eliminates the last part of the cycle marks a step forward.

Elida Karaivanova, business leader on the advanced manufacturing team at Arup (whose engineers helped design Ocado’s robot-bearing grids), thinks growing public concern about climate change makes such change inevitable. “No big organisation will be able to stand in front of its clients and not talk about sustainability,” she says.

Combined with other advanced production techniques — in particular additive manufacturing, or 3D printing — AI can help realise two of the key tenets of the circular economy: reduction of waste, and keeping products and materials in use.

Machines with worn-out parts that are no longer made do not have to be cast aside: replacements can be made with 3D printing. Since AI can help predict wear and tear, these replacements can be ready in good time. Lian Jye Su, principal analyst at ABI Research, points out that “DIY enthusiasts” can take this one step further and use the technology to improve their old machines — a process of optimisation that AI readily lends itself to.

Machine learning can also be used to suggest new and superior materials. For example, Zymergen, a California-based synthetic biology start-up, is applying it to microbial genomes in order to discover substances with novel properties; these can then be made on an industrial scale without resort to petrochemicals.

Professor Fiona Charnley, who works in the Centre for Circular Economy at the University of Exeter, says there are “many opportunities” to use AI in manufacturing. She is interested in how it can be used in the later parts of a product’s life cycle, for example with “product passports”, an idea advocated by the EU’s European Resource Efficiency Platform.

This is the idea that each product should have a code that can be scanned to access information on how it can be disassembled and its parts reused; as with other aspects of the so-called “internet of things”, the resulting flows of materials may benefit from AI-based management.

Meanwhile, Mr Su points out that, at the other end of the product lifecycle, generative design is “increasingly enhanced by AI and will contribute to sustainability in manufacturing processes”. With this technology, the user lays out goals for the design of their product, such as being lightweight or costing a certain amount to make, and the software presents them with every option. This can bring obvious benefits in terms of reducing use of materials.

“The use of AI and digital technology is imperative if we are going to accelerate the transition towards a circular economy,” Prof Charnley says.

There is a big financial upside too: according to Made Smarter, an industry-led review published by the UK government in 2017, the total economic benefit of using AI in manufacturing will be £8.6bn yearly and £74bn cumulatively by 2035. Manufacturers that use AI, it says, will see plant maintenance costs falling by 15-25 per cent, machine downtime reducing by 20-35 per cent and productivity increasing by 3-5 per cent.

Yet there are dangers. Some worry that AI-driven efficiencies could simply be used to accelerate unsustainable practices — in sectors such as “fast fashion”, say, or smartphones, which quickly become outdated.

Prof Charnley agrees that this is a risk as much of the focus in AI is on faster production. “We just need to start thinking about how we can use existing and state of the art technologies from a circular perspective,” she says.

Overall, Mr Su is optimistic. Whether or not manufacturers have sustainability in mind, by using AI to boost efficiency they will move towards a more circular economy anyway. “The whole [of] industry is becoming more efficient with how they use resources,” he says.

Copyright The Financial Times Limited 2023. All rights reserved.
Reuse this content (opens in new window) CommentsJump to comments section

Follow the topics in this article