A Cure for AI “Pilotitis”

Many organizations beginning to use Artificial Intelligence (AI) fall prey to “pilotitis,” a perilous state in which an enterprise relentlessly invests in ideas, but not in the discipline to kill those that are not working or in the art of scaling them and moving to production.

AI can digest and analyze enormous amounts of data much more quickly than a human could, and further enables computers to make decisions that were previously reserved for human minds.  AI cannot only act without human intervention but can also draw conclusions from the large amounts of data fed into it.  This transformation affects industries in which number and data are the main focus, such as accounting and finance.  As organizations begin their AI journeys, they must first recognize that data will be the driver. The richer the data, the better the results – whether video, audio or written material.

So, how do you start with AI? Chris Aiken, Executive Director in Advisory Services at Ernst & Young, provided a presentation to the FEI Committee on Finance and Information Technology at their June 2019 Meeting in Los Angeles, CA.  According to Aiken, organizations need to develop an enterprise strategy to create a shared vision, shared priorities, shared resources and shared accountability for results.  Here’s a ten-step plan:

  1. Define targeted business outcomes and human experiences
  2. Document capabilities, inventory data assets, establish performance baseline
  3. Holistically identify AI-driven automation and analytics use-cases applying multiple tenses
  4. Assess organizational and technical readiness to inform planning and priorities
  5. Identify vendors and pilots to support learning
  6. Architect AI systems from the full spectrum of technologies, tools and methods
  7. Consider impacts on people, including customers, vendors and employees
  8. Review operational, workforce, reputational and macro risks to fully inform stakeholders
  9. Plan for continuous discovery, skill-building and iterative training of AI systems
  10. Define realistic expectations for competitive advantage, investment and returns