How would human-centered design make AI effective?

When designing AI products or services, it’s tempting to assume that the best thing you can do is automate complex, repetitive, boring, or even dangerous tasks to do manually. However, there are many situations where human involvement is required, such as when people enjoy doing, when the stakes of the situation are high for their role, or where people don’t agree on the ”correct” way AI does. Hence, AI applications must reflect realistic conceptions of user needs and human psychology.

A major report on this conversion has been released by Deloitte. The report identifies where AI is at today and why AI needs human-centred design. Here are key takeaways from the report, which you can read in full here.

• Smarter together: Why artificial intelligence needs human-centered design (2018), Deloitte Review, Issue 22

Key aspects of Human-Centered AI Design: How does human-centered design make AI effective?

• Goal-relevance. Data science products and AI applications are most valuable when insightfully designed to satisfy the needs of human end users.
• Handoff. Human-centered design is needed to ensure any “handoff” moments from AI to human in exceptional or ambiguous situations that require common sense or contextual understanding.
• Feedback loops. Automated algorithmic decisions can reflect and amplify undesirable patterns in the data they are trained on user behavior and societal biases encoded in data. AI can use feedback loops to fix errors and learn from them.
• Psychological impact. Algorithms designed without a focus on their users’ human nature can impair user behavior.