- After over 2 years of implementing AI at Harrison Manufacturing, CEO Julie Harrison poses a core question: when using AI, which part of “intelligence” is handed to the machine and which part must humans retain. Real-world experience shows AI performs best in highly specific tasks, while many grand claims remain mere marketing.
- The company produces various lubricants and additives with extremely precise formulas. AI helps optimize production volume, speed, and accuracy within complex systems requiring strict control of temperature and pressure.
- Starting with Microsoft Copilot, the company first applied AI in the back office, seeing clear efficiency in finance, supply chain, and administrative tasks like debt management.
- A key principle learned is “garbage in, garbage out.” If input data is poor, AI yields flawed results. Thus, the entire production value chain must be redesigned for data standardization.
- Legacy systems can create bottlenecks when AI processes data faster than old machinery can respond, forcing businesses to synchronize technology.
- AI applications on production lines require human training and verification; this cannot be fully automated as it demands human curiosity, skepticism, and experience.
- Experienced engineers remain vital; they can sense issues through intuition and wisdom—invaluable traits that AI lacks.
- AI does not replace employees but boosts productivity. The best results occur when experts define AI’s role in specialized analyses.
- Conversely, the worst outcomes happen when employees become over-reliant on AI and lose vigilance. In manufacturing, one small error can ruin an entire batch.
- AI also opens doors in R&D, allowing the company to run scenarios and equations much faster than traditional methods using internal data. 📌 Conclusion: After 2 years, Harrison Manufacturing finds AI highly effective for specific tasks like process optimization and finance. However, success depends entirely on data quality and human oversight. AI accelerates speed and R&D but cannot replace human experience and skepticism. Core lesson: AI is a tool; the final decision remains human.
