As artificial intelligence rapidly moves from the digital world into the physical world, the manufacturing industry stands at a critical crossroads. For Honeywell, industrial AI has never been a matter of "whether to implement it or not," but rather a long-term challenge: how to implement it safely, reliably, and responsibly. Embracing innovation and maintaining prudence are not contradictory. Truly mature industrial intelligence does not pursue overnight disruption, but rather the gradual layering of intelligent capabilities on top of existing automation and engineering systems, continuously releasing value only after reliability, safety, and traceability have been fully verified.

Recently, the Global Science and Technology Innovation Conference and the 2025 Forbes China Science and Technology Innovation Figures Awards Ceremony were held in Beijing. In the "Physical AI, Future Factory" summit dialogue, Yu Feng, President of Honeywell Greater China, shared Honeywell's practical reflections and industry insights on the essence, application boundaries, responsibility bottom line, and value of humanoid robots in industrial AI.

I. The Difference in Capabilities: AI is Not "Smarter Automation"

Yu Feng first clarified the fundamental difference between industrial AI and traditional automation: In the era of automation, the core was to transform human experience and knowledge into executable engineering rules, and then use systems such as DCS, PLC, and MES to enable large industrial plants to achieve orderly and stable repetitive operation. In the era of intelligence, however, AI operates within a closed loop of perception, judgment, execution, and continuous learning—it can discover patterns that are difficult for humans to grasp from massive amounts of real-time and historical data, providing optimization suggestions that go beyond fixed rules.

But he immediately emphasized: "In process industries such as chemical and oil refining, a temperature deviation or a fluctuation in composition can trigger major safety and production accidents. Therefore, the core of industrial AI is never 'how smart' it is, but reliability. Intelligence divorced from safety and stability is meaningless in industrial scenarios." This is precisely the underlying logic that Honeywell has always adhered to: industrial AI must be based on industrial mechanisms and engineering experience, not a purely data-driven black-box model.

II. Boundaries and Responsibilities: AI Cannot Be a One-Shot Solution, Nor a Disruptive Revolution

Faced with the heated debate about whether AI will completely disrupt factories and achieve unmanned autonomous control, Yu Feng gave a clear answer: Industrial AI is a gradual evolution, not an overnight disruption. Especially in process industries such as oil refining, chemicals, and steel, which are crucial to national welfare and involve massive assets, safety and stability are inviolable bottom lines.

Honeywell's approach is to "overlay intelligence" on existing automation architectures, rather than starting from scratch. The recently commercialized Experion® intelligent predictive solution is a practical application of this concept—based on Honeywell's flagship Experion® PKS distributed control system, it integrates decades of industry mechanistic models with AI technology, mines the value of historical data, identifies anomalies in advance, predicts risks, and provides actionable operational suggestions, helping companies shift from passive response to proactive prediction and leap towards autonomous operation. It has already been piloted and its value validated by global energy companies such as Chevron and Total Energy.

Yu Feng further pointed out that there are two insurmountable boundaries for the implementation of industrial AI: First, the boundary of safety responsibility. Key control points must retain human intervention and final decision-making power; AI should always be a "super co-pilot," not a "driver" replacing humans. Second, the boundary of data and knowledge accumulation. Industrial experience and process lessons are scattered throughout equipment and enterprises. Data cleaning, knowledge accumulation, and model verification require a long period, far exceeding what can be completed in a few months. For large industrial enterprises, reliability, safety, and traceability are always more important than "being smarter."

III. The Debate on Humanoid Robots: Value Lies in Discreteness, Not Processes

Regarding the current hot topic of humanoid robots, Yu Feng, starting from the essence of industrial scenarios, offers a rational judgment: In the two major areas of industrial separation and process, the application value of humanoid robots in process industries is not prominent at this stage.

Process industries (chemical, steel, cement, smelting, etc.) are centered on continuous chemical reactions, parameter adjustment, and intelligent control. Production relies on the precise stability of parameters such as temperature, pressure, and flow rate, rather than limb manipulation. The robotic arms and mobility of humanoid robots are more suitable for assembly and handling scenarios in discrete manufacturing. Honeywell's focus has always been on physical AI in the process industries—enabling AI to interact directly with underlying control systems to optimize parameters, predict anomalies, and ensure safe and efficient continuous production, rather than chasing the short-term hype surrounding humanoid robots.

Conclusion: Responsible Industrial AI is the Key to Long-Term Value

The future of industrial AI lies not in radical disruption, but in responsible integration. Honeywell consistently adheres to the principles of industrial mechanisms, prioritizes safety and reliability, and follows a gradual evolutionary path, ensuring that AI becomes a tool for improving efficiency, ensuring safety, and passing on experience, rather than a source of risk.

For manufacturing, true intelligent transformation is not about piling up technologies, but about enabling AI to deeply collaborate with automation, people, and processes, while upholding boundaries and taking responsibility, leading to a safer, more efficient, and more sustainable future.

Recommended Products

Contact Us

Manager: Leonia

Email:sales11@amikon.cn

Whatsapp: +8618030175807