April 20, 2021
Artificial intelligence (AI) has the potential to revolutionise the manufacturing sector. AI helps manufacturers reduce waste and develop more efficient processes. However, AI is unlikely to provide satisfactory outcomes without a focus on the human end users.
Here is a closer look at the importance of a human-centred AI approach in manufacturing.
What Is Human-Centred AI?
A human-centred design involves the development of AI systems that collaborate with humans. With a human-centred approach, AI systems are set up on the existing data and human input is included where either data is not available or more creative problem-solving skills are needed. Human-centred AI also aims to deliver more efficient solutions by focusing more on the needs of users.
This approach helps create systems that support human activities instead of replacing them. A human-centred AI system should be more user friendly. Many companies mistakenly believe that AI provides a replacement for human workers.
AI is most effective when it enhances workers and business outcomes. AI can handle specific tasks, which gives humans more time to focus on work that requires human input.
Why Does Smart Manufacturing Require Human-Centred AI?
Many processes rely on more than just data to achieve optimal results. A successful AI solution needs to understand what humans want. Without human input, AI systems are often left with incomplete datasets, which increases the risk of failure.
AI systems do not possess certain human traits that are typically necessary for decision-making. An AI system does not have empathy or common-sense knowledge of social and cultural issues. It may lack the context needed for optimal outcomes. This becomes more of an issue with consumer-facing AI systems.
Humans are prone to making mistakes. However, AI systems may produce mistakes when they do not possess all the information needed to make effective decisions. In the manufacturing industry, a lack of human-centred design may impact the reliability and efficiency of an AI system.
Human-Centred AI Increases Reliability
Human input is needed to help make AI systems smarter and more reliable. Numbers and statistics do not always provide a complete picture. Algorithms lack the full capabilities of humans to consider the nuances of each process.
Manufacturing involves processes that benefit from human intervention. Ensuring that AI-powered tasks are designed with humans in mind increases the reliability of those tasks. For example, predictive maintenance is a common use of AI in manufacturing. With predictive maintenance, AI systems automate work orders, schedule tasks, and use sensors to gather data related to the performance of machines.
Algorithms analyse current and historical data to determine the best time to take a machine offline for maintenance. However, the AI system may not account for every factor that influences the decision of when to perform maintenance.
An increase in employee call-offs, changes to orders, increased or decreased demand, and other issues may impact the maintenance schedule. Human input ensures that the AI system understands these concerns and learns from them.
Human-Centred AI Improves Decision-Making
Along with making AI systems smarter, a human-centred design can help make humans smarter. Combining human input with an AI system allows humans to make smarter decisions. AI systems and humans each have their strengths when it comes to decision-making. AI systems have the computational power to analyse large sets of data. Humans can add context to the data analysis to inform their decisions.
AI systems that incorporate machine learning (ML) can learn to make more effective decisions. However, human input is still needed to give the ML software more perspective for understanding what humans want or need. Human-centred AI should enhance the way that workers make decisions instead of replacing the need for their input.
Human decision-making that is supported by AI systems also allows manufacturers to better address the needs of their customers. Humans have the empathy and common-sense needed to understand the pain points and concerns of their customer base.
The best AI solutions leverage human knowledge and complex data to understand what drives customer behaviours. AI systems can analyse customer behaviour to detect patterns while humans contextualise the information to enhance their decision-making process.
Human-Centred AI Improves Efficiency
As human-centred design focuses on enhancing human capabilities instead of replacing them, each AI-powered task involves a handoff. The efficiency of the handoff influences the overall efficiency of manufacturing operations. An ineffective design may also result in delays and errors.
Many AI systems run on autopilot until they require human input. AI systems that require human input should offer an efficient way for humans to provide input. For example, a sensor may monitor the temperature of a machine and alert a worker when the temperature exceeds a set threshold. A human-centred approach considers the steps that the worker needs to take after receiving the alert. If the process of responding to an alert is too complex, the response time is delayed.
One study found that 29% of AI systems in manufacturing are used for maintaining machinery and production equipment. The system may determine the best time to conduct maintenance but still needs a human to carry out the maintenance. The human-centred design considers the steps involved in carrying out the maintenance instead of focusing solely on when to complete it.
A human-centred approach to AI is also more scalable as scaling AI solutions requires human involvement. Workers may need to be trained on new technologies or processes. If these issues are already addressed in the initial design of the AI system, scaling the system becomes more efficient.
Conclusion
Human-centred AI combines the analytical power of the AI with the creative problem solving and experience of humans, effectively bringing together the best from both sides. Consumer-facing AI systems especially need to include a human-centred approach to avoid feedback loops and inaccurate results. However, the manufacturing industry also benefits from human-centred AI.
AI systems do not always have the context needed to make informed decisions. Incorporating human input in smart manufacturing processes helps increase the reliability and efficiency of AI systems. In the end, a human-centred approach helps AI systems and humans become smarter. They support and learn from each other to complete complex tasks.
