Applying AI and Analytics to Emissions Data

There’s little doubt about the impact of AI on businesses. When it comes to an organization’s emissions data, it is generally understood that applying AI should bring useful insights from this critical data.

In terms of emissions data, companies are challenged in two areas. The first is driven by regulations requiring emissions reporting and ultimately emissions reduction over time. The second area relates to spotting business opportunities related to emissions such as investment credits available from the U.S. Inflation Reduction Act for carbon sequestration and storage.

Using AI against large emissions datasets can highlight opportunities for emissions reduction relating to specific parts of a business or supply chain or opportunities for investment related to emissions.

A prerequisite is that the data must be well organized and consistent. Consistency includes data and metadata for emissions activity, units of measure, the emissions calculation formula used and the emissions component category.  The data consistency issue extends to organizational attributes describing the structure of the company's organizational boundaries, location, facility and equipment.  It also extends to descriptions of product life cycles and the reporting domain.

Related:Google, Microsoft, OpenAI Square Up on Generative AI

Emissions data is generated from many disparate data sources to the extent that the data cannot be standardized and AI model performance will suffer. Despite various greenhouse gas data standards, once data is captured in internal IT systems, the lack of a ubiquitous data model standard means data collected from various parts of a business may use different data naming and units of measure and this is especially true for Scope 3 emissions data collected from suppliers and business partners. In addition, whatever data integration approaches are used, a consistent data framework will be a requirement before applying AI to emissions data in a complex enterprise.

One possible approach is to adopt a standard data model like the Open Footprint data model within the enterprise and throughout the supply chain, ensuring consistency in the data naming, metadata, units of measure and relationships between data elements.

For example, a large multinational corporation would like to utilize AI queries to understand both the emissions profiles of various suppliers and which suppliers are reducing emissions effectively over time. Understanding supplier emissions and changes over time can inform sourcing decisions, meaningfully impacting the multi-national company’s Scope 3 emissions. Therefore, emissions data from their suppliers has to be consistent in terms of data and metadata definitions.

Related:Address Bias in Generative AI to Unlock Benefits While Minimizing Risk

Another example would be a company trying to assess Scopes 1 and 2 emissions from across their business, looking for areas where capital investment projects can best result in reductions in emissions, requiring that the Scope 1 and 2 emissions be developed using consistent data definitions, units of measure and calculation methods so that data is comparable across disparate businesses.  This extends to Scope 3 data from supply chains that may be different depending on which business unit is being evaluated.

Adopting a common emissions data model has numerous benefits, one of which is that it will make gathering useful data from your organization's value chain and throughout your organization easier.  The main reasons may be to help facilitate the use of AI to provide advanced analytics and unlock the business value inherent in emissions data.