The power of AI comes with a lot of responsibilities. It has changed a lot about our world, but the change has been a double-edged sword. It is responsible for a lot of progress, but some risks and unknowns are involved in its use. Because of this duality, most business leaders are unsure how to use and deploy AI systems. Still, they can look to public agencies that have already experienced these uncertainties in AI initiatives for some answers.

The 5 Major Areas of AI Uncertainty

Since 2014, Queensland University of Technology has engaged in around 40 projects that deal with AI efforts in the public, private, and nonprofit sectors in Australasia and the United States. However, most of the recommendations were drawn from the public sector. This experience allowed them to observe managers’ use of AI firsthand, and they found five major areas of uncertainty.

1. Communicating Realistic Expectations

From these projects, the team found that managers weren’t sure about the outcome of AI initiatives, and there were four main ways they thought about AI:

• AI is overhyped
• AI can fully automate tasks
• No opinion
• AI isn’t perfect, but it is getting better every day

The last group has the right idea. Skill training and accountability were the best way to gain support from the first three groups. This education needs to begin with communicating what AI can accomplish and what it can’t.

Want More Tech News? Subscribe to ComputingEdge Newsletter Today!

2. Assessing How AI Fits Tasks

Managers struggled with evaluating how AI fits into their business. It is the degree of discretion that determines how to fit AI into a task. Repetitive, low-level tasks like data entry or facilities operation are examples of a low-discretion take suitable for total automation. For tasks that require medium discretion, like developing a hiring process, AI can augment human decision-making. But AI should not make highly sensitive and complex decisions.

3. Clarifying a Path for Deployment