A version of this article appeared in the NACD Directorship publication (https://read.nxtbook.com/nacd/directorship/march_april_2020/in_practice.html)

As a board member, you’re likely aware of the many strategic opportunities and threats posed by AI. According to IDC, spending on AI hardware and software is increasing at a CAGR of 24%. AI-driven projects will rapidly become a substantial percentage of any company’s investment in technology. 

 

What you may not be aware of are the environmental impacts of AI. AI can contribute to your company’s carbon footprint or if managed well, help reduce the impact your company has on the environment.

 

The issue of environmental sustainability is becoming a more critical issue to CEOs and boards as the finance world drives this change. Recently, BlackRock’s CEO announced that his firm now has a core goal of investing with environmental sustainability in mind. Goldman Sachs has now made “sustainable finance” core to its business. 

 

Additionally, at the Davos World Economic Forum, the International Business Council announced metrics and a framework for reporting on subjects that include environmental sustainability goals. The Big Four accounting groups an many of the world’s largest companies have signed up to account for (among other things) environmental impact. 

 

As a board member, it’s becoming increasingly important to understand your company’s environmental impact. As you will see, AI can play an important role. In this article, we’ll explore the environmental impact of AI and how boards should approach oversight of this increasingly important issue. 

 

AI Can Have A Substantial Carbon Footprint

 

AI-based systems are highly compute-intensive. They must process a great deal of data, expanding the need for servers and dependence on energy to cool data centers. The adoption of AI within a corporation will increase the company’s use of energy. 

 

According to one study by the University of Massachusetts, training AI models to do Natural Language Processing (NLP), can produce the carbon dioxide equivalent of 5X the lifetime emissions of the American car, or the equivalent of 300 round-trip flights between San Francisco and New York.

 

Source: SP Global   https://www.spglobal.com/marketintelligence/en/news-insights/trending/HyvwuXMO9YgqHfj7J6tGlA2

 

This example of CO2 emissions from AI is stunning and disturbing. It is a wake-up call for us all. However, before we put too much emphasis on these results, we need to look at the bigger picture.

 

This is only one study for a specific type of AI which is not commonly used. More representative training tasks produce relatively tiny amounts of carbon. However, just because the most common methods of AI today aren’t significant carbon producers, doesn’t mean they might not become serious contributors in the future. At this moment, few studies exist that help a company evaluate the carbon impact of AI. 

 

What this study tells us is that we need to know a great deal more about the carbon footprint of all types of AI. The issue today is that we don’t know. We might be producing a little; we might be producing a great deal, but we need to know. It is part of the board’s responsibility to ensure we understand what is happening.

 

As board members, we need to ask for the information that is not being collected at this time by your AI teams. Without data about your AI carbon footprint, you may be creating a surprise reputational risk for the company once information about your carbon footprint becomes public.

 

Without data about the potential environmental impact of your future AI projects, you cannot fully evaluate your investment in any project. Now more than ever, the impact of AI on CO2 emissions needs to be a key element in your decision-making process.

 

Interestingly, Canada’s Montreal Institute for Learning Algorithms has recently released a tool designed to estimate how much carbon is produced in training machine learning models. This particular tool is a small step in the right direction. At this moment, few other tools exist. It will be incumbent upon your AI teams to either use other tools or create their own to address the board’s questions about environmental impact.

 

As environmental sustainability becomes more important, boards need a lot more information about the impact their company is having to provide oversight. Boards need to encourage senior management to track and report on what is happening inside their company as it relates to environmental sustainability. AI has the potential to produce significant and impactful carbon emissions. AI also has the potential to offset or reduce those carbon emissions.

 

AI Can Be Used To Reduce Carbon Footprint

 

Companies should consider allying themselves with any cloud provider that is committed to reducing their carbon footprint, thereby reducing their own. Instead of focusing on major internal projects to reduce environmental impact, it’s possible to shift a company’s AI training and processing to a data center cloud provider that can do that for you. For example:

 

Google’s DeepMind division has developed AI that teaches itself to minimize the use of energy to cool Google’s data centers. As a result, Google reduced its data center energy requirements by 35%. Google’s public cloud offering is called Google Cloud Platform.

 

Microsoft has committed to be carbon negative by 2030. Microsoft also runs massive public data centers (cloud offerings) under the name Microsoft Azure.

 

Amazon has a long-term goal of powering its global infrastructure using 100% renewable energy. This includes its cloud platform AWS.

 

AI can be a net positive contributor to environmental sustainability in many industries. Here are some examples:

 

In agriculture, AI can transform production by better monitoring and managing environmental conditions and crop yields. AI can help reduce both fertilizer and water, all while improving crop yields.

 

In energy, AI can use deep predictive capabilities and intelligent grid systems to manage the demand and supply of renewable energy. By more accurately predicting weather patterns, AI can optimize efficiency, cutting costs, and unnecessary carbon pollution generation.

 

In transportation, AI can help reduce traffic congestion, improve the transport of cargo (supply chain logistics), and enable more and more autonomous driving capability. AI will eventually help with the “last mile” delivery problem and reduce the need for delivery vehicles. AI can help businesses with demand forecasting, helping to reduce the amount of transport needed.

 

In water resource management, AI can help reduce or eliminate waste while lowering costs and lessening environmental impact. AI-driven localized weather forecasting will help reduce water usage.

 

In manufacturing, AI can help reduce waste and energy use in production facilities. Robotics can enable better precision. AI can design more efficient systems.

 

In facilities management, AI can help recycle heat within buildings and maximize the efficiency of heating and cooling. AI can help optimize energy use in buildings by tracking the number of people in a room or predicting the availability of renewable energy sources.

 

In materials science, AI can help researchers find new materials for solar panels, for turning heat back into useful electricity and to help find absorbent materials as components of CO2 scrubbers (taking CO2 out of the atmosphere.)

 

In AI itself, AI can be used to eliminate useless or redundant processing, thus reducing the computations that require so much power.

 

As you consider your industry, think about three areas where AI is likely to have an impact on environmental sustainability:

 

  1. Error reduction. When humans make errors conducting manual tasks, that work often has to be reviewed and re-done. The effect of addressing these avoidable problems is more energy use. AI can be a factor in reducing human error in many tasks. 
  2. Greater efficiency. By combining several types of AI, including machine learning, NLP, and computer vision, a company can create more efficient processes, reducing energy use. Also, AI can be used to remove unnecessary steps from the current process by contributing to the re-engineering of processes.
  3. Raw materials. When AI is focused on monitoring raw materials use, it can create opportunities to use less. AI can also be used to drive the creation of low-carbon materials for your products.

 

AI uses a great deal of energy, and most companies have no idea how to measure environmental impact. Board members should drive the discussion around the awareness and measurement of AI’s impact on the environment. 

 

Also, board members should further educate themselves on the opportunities enabled by AI to address the issue of environmental sustainability.

 

Glenn Gow is a Board Member and CEO Coach with an expertise in AI strategy. Contact him at LinkedIn.com/in/glenngow.