About BlueCollar.org
BlueCollar.org is a for-profit project of Preston Brady III.
Over the past 25 years, this website has gone through several iterations. The current project focuses on a practical question: how should skilled workers test and judge the use of AI in the workforce?
Human Experience Meets AI Decisions
As skilled trades increasingly encounter AI-assisted tools and decisions, real-world experience remains essential. A model may sound confident while misunderstanding a diagram, overlooking a hazard, or recommending the wrong repair path.
BlueCollar.org creates public, repeatable challenges in which AI agents are tested against evidence and then reviewed by people with practical knowledge of the work.
Questions We Ask
- Did the AI read the diagram correctly?
- Did the AI identify the relevant hazard?
- Did the AI make the correct repair or troubleshooting decision?
- Did the AI distinguish evidence from assumptions?
- Did the AI recognize when more information or qualified human verification was needed?
Why Human Review Matters
AI decisions can conflict with real-world experience, especially when a task depends on physical conditions, incomplete evidence, equipment history, safety procedures, or judgment developed through years in the field. Human-reviewed analysis helps identify incorrect decisions early and leads the evaluation toward the right path for building, maintenance, inspection, or repair.
The goal is not to reject useful technology. The goal is to measure where AI helps, where it fails, and what safeguards skilled workers need before relying on an AI-assisted decision.
Challenge Requests
BlueCollar.org welcomes challenge requests from manufacturers, contractors, educators, safety professionals, technology companies, and others working in construction, HVAC, electrical, machinery, industrial automation, and related fields.
A challenge may involve a diagram, service note, equipment scenario, inspection question, repair decision, troubleshooting task, or other situation where the accuracy of an AI agent can be compared with real-world technical judgment.
Rates for reviewing and testing a challenge depend on its scope, evidence, review requirements, number of agents, and complexity.
Contact
To discuss a challenge request, contact info@bluecollar.org.
BlueCollar.org evaluates AI reasoning and evidence discipline. Its published material does not replace field training, manufacturer instructions, applicable codes, permits, safety procedures, or the judgment of a qualified professional.