AI can help people do more. People decide what the work is for and how to check whether it is helping. They remain responsible for the consequences of using AI. Hi on AI gathers our research and published opinions, along with conversations about how to put that responsibility into practice.
Read Carl McQueen’s published opinions and hear his conversations with WPHM and WKZO about how people can have a say in decisions involving AI.
NEW
Radio interview ·
WKZO · Kalamazoo Mornings with Ken Lanphear · 8 minutes
What communities need to know before a data-center decision
Carl joined Ken Lanphear on WKZO to discuss the questions communities need answered about a data-center proposal. He recommends explaining early which terms can change and which are fixed. People need a voice in the decision and a way to check whether promised benefits follow. The conversation applies ideas from our research into people’s expectations of AI. That research does not establish support for any particular data-center project.
Crain’s Grand Rapids Business + Crain’s Detroit Business
Future of AI is not on the extremes, but Americans want to have a say
People can find AI useful and still want control over decisions that affect them. Carl’s Crain’s op-ed examines the conditions respondents placed on its use. They wanted a human option for important decisions and clearer rules at work and school. They also wanted opportunities to develop their skills. He argues that leaders should pay closer attention to those expectations as they decide how AI will be used. The same column appeared in both Crain’s editions.
Carl joins Paul Miller to discuss how people use AI and who answers for mistakes. A hiring example makes the concern concrete: an applicant needs a way to correct bad information and reach someone who can reconsider a decision. The conversation also covers learning at work and what regular AI users expect. Listen to the full interview on WPHM.
Carl’s Detroit News column applies the AI research to the debate over data centers. He recommends listening to the people affected, addressing benefits and concerns together, offering meaningful choices, and making a person with authority answerable for the project. Customers, neighbors, and ratepayers can have different interests. The column argues for studying those interests directly and showing how the proposed benefits will be checked.
These proposed McQueen Analytics positions describe what we would ask leaders to address when deciding how to use AI.
01
Give human review the power to change a decision.
A person affected by an important AI decision should be able to reach someone who understands it and can act. That reviewer needs the authority to correct information and reconsider the outcome. They should also be able to explain what happened.
02
Show where AI improves the work.
Choose a specific task and decide how to check the result. Measure the quality of the work and its consequences for the people using it. Check whether claimed time savings survive the work needed to correct errors. Ask customers and staff about their experience.
03
Give people a say in how AI affects them.
Explain where AI is used and what it is allowed to do. Listen to the people affected, including those whose information is used to make the decision. Make the available choices clear, including how someone can raise a concern or request review.
04
Protect the opportunity to learn.
AI can offer feedback and help people attempt unfamiliar work. Employers still need to give new staff the practice that builds judgment. Someone who checks an AI answer needs enough experience to recognize when it is wrong.
05
Name the people responsible.
Identify who approves a use of AI and who checks the work. Make it clear who responds when AI use causes a problem. People need a way to reach someone who can correct an error. Responsibility should remain visible as the technology changes.
06
Listen before committing a community to a project.
Ask people about the benefits they expect and the costs or concerns they face. For a data center, a neighbor’s concerns may differ from those of a customer. Ratepayers need answers about shared costs. Explain early which terms can change and which are fixed. Show how people can influence the proposal. Name who will check the promised benefits after decisions are made and who has authority to respond if they fall short.
07
Keep claims within the evidence.
Surveys can describe respondents’ reported behavior and views. Those answers do not establish future job losses. They also do not prove that an AI system is safe or guarantee support for a local project. Recommendations should state what the evidence supports and what still needs to be learned.
Hi on AI · perspectives
Carry the question into your work
These companions connect the published pieces to decisions leaders need to make.
If your organization is deciding how to use AI, bring us the question and what you need to understand before choosing a direction. Please leave out personal or confidential case details.
