Hardly a day goes by when we don't see an article on how we are likely to miss the Paris targets for climate emissions reductions. With 2020 just around the corner, the milestone for bending the curve on greenhouse gas emissions seems harder and harder to achieve. Indications of what might lie in store are all around. Let's imagine the clock rolled forward to 2050. What might a person born today and early in their career at mid-century ask the business and industry leaders of today?

Given the scale of what is unfolding, it would not be unreasonable for that young person to ask 'What were you thinking?' and 'Why didn't you take action when you could have, when you were in a position to do so?' Of course there are many who had been very active, but by far the majority did little to combat climate change. Jeremy Lent has suggested that the problem lies with patterning instincts, causing behaviours to be locked in so that is nigh impossible to make meaningful change.

Some might argue that the scale of the problem is so huge and the solutions are not available. Yes, the problem is huge but solutions are available. Many have been documented through Project Drawdown, for example; and there are many others. Perhaps the real problem lays in the way that business and industry is managed? What if algorithms designed to run enterprise, deliver superior business results and reduce greenhouse gas emissions were used to mobilise a new approach to climate change? To transcend the inertia and guide leaders and senior managers into new ways of making decisions that are vastly superior to current approaches and which shift us onto the trajectory that the world agreed to under the Paris Treaty. Let's call those algorithms and the myriad technologies and techniques that go with them Green Artificial Intelligence or Green AI.

'We can't run business by algorithms!' might be a first response. Well, we already do; they are built into the financial trading systems that control the world's financial systems and guide the billions of transactions that have been occurring while you read this. Those algorithms work with all sorts of checks and balances and operate under a myriad of regulatory, supervisory and other constraints, the most of important of which is often to make a profit on transactions. What if those trading algorithms were embellished with AI that guided business to reduce unnecessary spend on energy, water and materials, to design environmentally friendly goods and services, and to avoid the Earth and its systems of natural capital going bankrupt? Saving on costs often goes with increasing profitability; indeed, Porter argued back in 1995 that 'green' enabled businesses to be more resource productive and innovative and therefore more competitive.

There are many areas in which Green AI can make an impact, and there are many examples of steps already being taken. Examples of large-scale Green AI programmes include Microsoft's AI for Earth programme and Google's Green Data Centre; these programmes make practical use of Green AI. The Microsoft initiative provides a map of projects in which Green AI are being funded.

Some categories of applications of Green AI are shown in the following table. These include (1) Resource Productivity, (2) New Business Models, (3) Inspired by Nature, (4) Enterprise Awareness and Intelligence and (5) Trust.

Categories of Application Examples
Resource productivity & Ecodesign
  • elimination & substitution (e.g. of materials in products)
  • reuse & remanufacturing
  • reduction, streamlining & optimisation
  • design of goods and services for reduced climate impact
New Business Models
  • transparency & visualisation (eg through supply chain)
  • value chain productivity (including digital value chains)
  • decision support (e.g. for smart cities)
  • consumer coalitions & collaborative journeys
  • organisational design (including Green Business Model Canvas)
  • environmental and resource monitoring/assessment
Inspired by Nature
  • circular enterprise
  • green chemistry
  • biomaterials
  • renewable energy systems
  • natural capital accounting
  • protection of life support systems
Enterprise Awareness and Intelligence
  • monitoring earth and climate systems
  • digital twins (proximal, remote, hybrid)
  • traceability & benefit tracking (e.g. through supply chains)
  • competitor intelligence & competitive strategy
  • trustworthy disclosure
  • distributed systems (including Green blockchain)
  • earth tokens


Table 1. Illustration of examples of application of Green AI.

Examples of high-impact initiatives can be identified. In an attempt to create a more transparent supply of palm oil and avoid deforestation in commodity supply chains, Unilever now publicly discloses a list of the suppliers and mills from which it directly and indirectly sources palm oil. The undertaking has involved mapping 300 direct suppliers and 1,400 mills, looking not just at where palm oil enters Unilever's supply chain, but the processors, middle men and agents along the chain. Other companies are also disclosing their suppliers. As another example, Walmart's Project Gigaton has the target of driving one billion tonnes of CO2e out of the company's global supply chain over about 140 months. Imagine what would happen if every company took that approach and the process was guided by Green AI to quickly identify the opportunities and act on them. Greenhouse gas emissions would nosedive as the global economy embraced a positive approach to stewardship of the earth.

The recent announcement by insurer Allianz that it is integrating environmental and sustainability considerations 'into the DNA of every employee' implies that business processes in that corporation are likely to reflect a company that is changing its relationship with the earth. The ripples are likely to spread into the many markets and professions that Allianz interacts and engages with. The work of brokers, underwriters, actuaries, loss adjusters, claims professionals and others in the sector will undoubtedly change. The DNA of insurance products, the legal language around which insurance contracts are written, will change, perhaps requiring businesses to demonstrate they have taken climate mitigation actions for their insurance cover to remain valid. AI is already used extensively across the insurance sector and there is the potential for Green AI to play a key role in development of Green Insurance and more broadly across the financial sector.

Satellite and drone imagery combined with Green AI is now being used to identify energy, water and materials usage in industrial sites without information from the sites. This allows insurers to rank sites according to energy, water and other performance metrics and guide risk-based pricing. New forms of actuarial indices are emerging which help the insurer to select and price risk. Satellites and drones are being used to identify sources of methane emissions on industrial sites. To identify water leaks in utility distribution systems. To track changes in forests and verify that timber has been sourced responsibly. To monitor construction sites to confirm that building work conforms to the planning consent. To monitor for flooding and predict how facilities will respond. A wide range of processes and activities is already creating a visual intelligence about our interaction with the earth. Underpinning many of those new systems are forms of Green AI. Whether it is pattern recognition to identify the shapes and sizes of buildings across a city to locate roof spaces for solar panels, pattern recognition to infer industrial operations, analytics to investigate pollution streams, infrared systems to detect heat losses from offices and data centres, the range of Green AI applications is growing daily. Some of these systems are beginning to integrate with climate change models that have been developed over recent decades and allow business scenarios to be evaluated under changing climate regimes.

Digital Twins are another example of where Green AI is being used to make businesses more productive. Patterns within data from a wealth of sensory systems that makes up the Internet of Things (IoT) can be detected using AI to design better processes and ensure more productive operations. The main focus of the Digital Twin in the industrial context is on asset utilisation, product definition and deep introspection inside the business using big data streams. Most Digital Twins are facilitated through sensors, actuators and other systems that connect directly with machinery. Others, such as Remote Digital Twins (RDT), use aerial or other imagery or data combined with AI-based analytics to look inside and outside the site/business to understand the enterprise in its changing landscape & how it can adapt. It provides insight into business growth, options for change and risk management and relies heavily on benchmarking and situation analysis. For example, the RDT allows an underwriter to run scenarios on an insured facility to test the impact of events such as utility blackouts, hurricanes, floods and drought. Digital Twins are being constructed also for cities, agricultural systems, and critical infrastructures.

For business leaders and investors who are looking to the future, Green AI helps gain insight into building new enterprises that have significantly lower costs and lower carbon profiles than competitors & their supply chains. Green AI helps in the design of new products that have environmental superiority. It is now possible to evaluate new enterprise configurations before they are built, much akin to 'genetic engineering' but on the scale of business building blocks for earth-sensitive enterprises. This is likely to be a critical area that Green AI can help business adapt to markets constrained in their GHG emissions. What were once supply chains consisting of largely disconnected silos can become highly integrated networks functioning with green machine intelligence. Green AI can be used by a company to analyse in detail the resource flows through competitors' systems. For example, we may see Green AI used to identify counterfeit and cloned products being placed on the market by competitors; such products often have inferior environmental credentials (as well infringing intellectual property and other rights).

By 2030 each entity in the supply chain may be fully transparent in terms of greenhouse gas emissions at the process step. This is likely to bring increased levels of R&D focussed on innovation to design and build enterprise systems that repair and regenerate natural systems. Styles of leadership are likely to be very different from those in current supply chains. By 2030 many of the key lock-downs will have been put in place to operate within planet-wide carbon-constrained enterprise logic. There will be greater understanding of the role of enterprise in engaging with an operating system aligned with nature, and striving for climate (and therefore public) safety. The role of 'licence to operate' will have transformed into new mechanisms, often guided, perhaps even controlled, by Green AI systems.

Green AI has the potential not only to deliver meaningful outcomes in emissions reduction but also to help build a green super-industry that through digital collaboration could effectively halt climate change. The companies that build that super-industry, and deliver the systems and services through an Earth Digital Twin, are the ones who will not only hold the new market positions but also will build trust amongst the public, at a time when public trust in traditional corporations is at an all-time low.

Some people argue that AI could bring about a scary future in which the machines take over and reveal little about their workings to humans. Perhaps. But is such a future more scary than a planet for which the climate change we are driving ensures ecocide and makes it is impossible for most species, including humans, to live? Within the context of climate change it may be interesting to reflect on Isaac Asimov's Laws of Robotics, the first of which was 'A robot may not injure a human being or, through inaction, allow a human being to come to harm'. Might the Earthbots operating within the Earth Digital Twin give us a new set of laws, the first of which might be 'A human being may not cause climate injury to another human being or, through climate inaction, allow another human being to come to harm'?

About the author

Michael Gell works at the nexus of energy, environment and digital technologies. His career began with British Gas R&D in the late 1970’s. Following a decade of research into industrial energy efficiency and low-energy houses, his focus broadened to advanced electronics with IBM and MOD. He established BT's research centre on quantum electronics in 1986.

In 1990 he set up the corporation's research centre on digital technologies which spearheaded the transformation of the telephone giant into the internet age. Michael established an independent energy and environmental business in 1995 and this has evolved to Greenclick. The company specialises in enterprise systems in the green economy. Specific areas of research include Green AI in the design of enterprises aligned with Nature. Michael is a regular keynote speaker.