Human and AI systems can bolster an organization’s leadership, teamwork, creativity and social skills while amplifying the company’s speed, scalability and quantitative capabilities. In order to do that, business leaders need a clear framework for:

• How employees can train and guide the AI.
• How AI can bring out the best in their workforce.
• How business processes should be redesigned to help human-AI collaboration.

Further, business leaders have to explore how human-AI systems can advance their enterprise goals. They have to evaluate:

• Opportunities for AI to work independently, streamlining operations.
• Where AI can collaborate with human colleagues to improve performance.

Managers need a best practices framework to design their human and AI collaboration by considering the nature of the collaboration as well as these characteristics:

• Transparency
• Trust
• Responsibility for specific tasks
• Levels of autonomy

The Partnership on AI organization outlines the four questions that business executives should ask before designing a human-AI collaboration framework within their enterprise²:

1. What is the nature of the collaboration?
   • Stage of development: is the AI fixed or will it continue to learn from interactions with humans; what will be the level of collaboration; who will work with the AI?
   • Goals of the collaboration: do humans and AI know the goal and nature of the collaboration; are human and AI goals aligned?
   • Pattern of collaboration: how long will the collaboration last, and will interactions between humans and AI happen in parallel, or will they take turns?
   • Level of autonomy: will people or AI contribute more to the decision-making, and how much autonomy does the AI have?
2. What is the context of the collaboration?
   • Location: how many people will interact with the AI; will they operate from the same location or virtually?
   • Awareness: is the human aware they’re interacting with AI, and do they need to give consent before interaction?
   • Consequences: what would happen if the AI failed to perform as expected; what are the benefits if the AI performs as expected; are there broader consequences of this collaboration; how are people factoring in privacy and security in the process?
   • Assessment: who is measuring the effectiveness of the collaboration, and are the measures subjective or objective?
   • Level of Trust: do the AI and humans trust each other and are both trustworthy?
3. What are the characteristics of the AI system?
   • Interactivity: how are people interacting with the AI (screen, voice, wearable, VR, etc.); could the nature of the data the AI is using influence the interaction?
   • Adaptability: is the AI only generating information or predicting the next step in the interaction process?
   • Performance: how predictable is the AI, and how often does it produce false positives or negatives?
   • Explainability: can the AI transparently communicate its decision-making process – including what inputs it used – to people?
   • Personification: is the AI human-like, and how easily can it be anthropomorphized?
4. What human characteristics has the system taken into account?
   • Age: will the age of the human affect the collaborating, including whether a child will use the system?
   • Differently-abled: are people with different needs able to use the system?
   • Culture: are there cultural norms for people working with AI?

For human-AI collaborations to reach their full potential, business leaders have to design systems that are useful, safe, transparent and beneficial. Taking these metrics into account will also help users trust the system and easily embed it in their operations.

Chief Scientist and Managing Director of US Government Accountability Office (GAO), Dr. Tim Persons says, “I think we’re still underestimating how much we’re going to get out of [AI] over time as it evolves. I think it’s going to surprise us.”¹