Optimizing AI Investments: How To Leverage Activity-Based Costing

Optimizing AI investments requires transparency and effective cost management, which can be achieved through activity-based costing (ABC) to ensure AI deployments deliver measurable value.

Data alone does not serve many useful purposes. Processed data, however, results in the creation of information and knowledge. In order to develop effective AI solutions, making objective, consistent, complete, comparable, relevant and timely data available to the AI models is crucial.

Irrespective of the sources of data and whether it is structured or unstructured, the cost and complexity of getting it usable by AI models should not be underestimated—nor should the cost of training these models effectively and weeding out bias.

Given the absence of “reasoning” from AI, it is hard to determine the accuracy of these models, and as such, considerable effort goes into the training and retraining process. When things go wrong, it is important to have the ability to trace back the validity and data sources.

Having the metadata and paradata available (paradata is information on how the data was collected), makes this task somewhat easier. The data readiness step in an AI deployment could, in my experience, account for over 70% of the total cost of deployment, hence it’s so important to get it right.

With all that said, it is equally important that the cost-benefit ratio of an AI deployment is not only understood but also significantly favorable. In this regard, I have found that the use of activity-based costing to evaluate AI and knowledge discovery system deployments is a good practice to adopt. AI is an investment, irrespective of whether the system is deployed at the operational or strategic levels of business.

Nevertheless, it should be subject to the same level of scrutiny as any other investment.[…]