Building Trust in AI: The Path to Impactful Automation

There’s no denying the transformational impact of AI within the manufacturing sector. It’s becoming increasingly clear, with a reported 35% of manufacturing firms using AI for purposes such as predictive maintenance, quality control, and demand forecasting. And it’s expected to be an ongoing trend, with AI projected to boost manufacturing productivity by 40% or more over the next decade. 

However, trust will be a key factor if such projections are to be realised. Indeed, the Harvard Business Review emphasises that “As AI becomes more powerful, it faces a major trust problem.” We’ll explore reasons for this problem and ways of building trust in AI here.

The Trust Deficit

There’s a well-established link between trust and the adoption of AI systems. This has been confirmed by an in-depth BSI report, revealing that 84% of organisations believe employee buy-in to be important for the successful roll-out of AI in business. However, the lack of trust continues to be a significant barrier, with a recent survey of UK manufacturers highlighting the adoption of AI processes by a mere 15%, with just 13% intending to actively use them. Three in five people are wary about trusting AI systems, with 67% having low or moderate trust in the technology, according to an international KPMG study.

Factors contributing to the lack of trust include:

• The lack of decision-making transparency and difficulty in validating AI-generated data
• Concerns over the safety risk of AI errors
• The likelihood of discrimination due to biased training data
• Worries over the displacement and replacement of human workers.

There’s an obvious need for a long-term approach, with the adoption and integration of AI being seen as more of a marathon than a sprint. This applies particularly in the UK, where there’s more of an inclination towards caution than in other nations. There’s also some variation in the levels of adoption and willingness to trust AI across different sectors.

Building Blocks of Trust

The varying levels of trust can be partly put down to the approaches of different governments, with the adoption of AI being highlighted as a strategic growth priority in countries such as India and China. In contrast, the introduction of legislation and safeguards can be seen as impacting the levels of trust across Europe. Of course, there are also concerns over the potential of job losses, highlighting the need for a well-thought-out and continuous approach to AI integration.

While reassurances should be given, there shouldn’t be any concerns over the collection, storage, and use of personal data in manufacturing environments. It’s also important to focus on the workforce opportunities enabled through AI, with the continued reliance on high-level human capabilities such as creative problem-solving. With the right tools, companies can create AI specific to their own environments and with domain-specific knowledge.

Overcoming Organisational Challenges

The Manifesto for Automation highlights the scale of the workforce challenge in the UK, with firms across all sectors of manufacturing competing for a small pool of talent. There’s clear scope for overcoming this challenge through the wider adoption of automation and AI. This will depend to some extent on the availability of financial support, with the “robots take jobs myth” also having to be busted. Investment in such technologies should be actively promoted, given the potential for automating jobs in the dull, dirty, and dangerous categories. Benefits such as improved productivity and efficiency should be emphasised, with AI enabling better customer service for business growth.

The BSI report revealed a shared belief in the importance of training for the safe, ethical and effective use of AI. 87% of business leaders said that such training would also enhance the protection of jobs. However, there’s more work to be done, with only a third of those leaders saying that such training was of a significant volume. Employees should also be better informed of how AI is being used for the realisation of business plans. Trust can be further enhanced through improved compliance and implementation of guidelines for the safe, ethical and responsible development of AI.

Other measures for trust in AI include:

• Developing clear lines of responsibility for AI outcomes
• Taking steps to address the harm caused by biased or erroneous decisions
• Conducting regular reviews to ensure the realisation of AI benefits.

Building a Foundation of Trust

The case for investment in AI technologies can be linked with case studies, like that of Ishida’s Sentinel 5.0. The integration of this AI-driven remote production monitoring and reporting software has enabled food producers and manufacturers to unlock benefits such as enhanced throughput, performance, and efficiency. With its machine-learning capacity, the Sentinel 5.0 indicates production line issues and highlights the need for preventative maintenance. So performance can be optimised and value driven with minimal need for human input.

The realisation of such benefits points to the need for the integration of AI as part of the wider business strategy. In taking such an approach, organisations can enhance their internal processes, deliver better customer service, and be more successful. Trust can be built over time with the safe, ethical and responsible implementation of AI according to international laws and guidelines. Transparency and collaboration will be key, with employees and other stakeholders kept informed of how and why AI is being used. This can drive the success of the UK’s manufacturing sector.

As the leading trade association for automation suppliers and technology end users, Automate UK is committed to unlocking and supporting the realisation of AI benefits. We will fulfil these aims through technical support and expertise, with our members benefiting from access to world-class exhibitions, subsidies, training, and much more besides. The upcoming forum on AI in Manufacturing is the perfect example, bringing together end users, academia, and product suppliers to discuss what AI looks like on the production line. Go ahead and book your tickets for this event, scheduled to be held on 22nd of July.