Harnessing the power of large language models for manufacturing

The manufacturing industry is continually striving to harness the latest technological breakthroughs in the relentless pursuit of improved automation, heightened operational transparency and faster product and technology development.

A paradigm-shifting advancement now rising to prominence within the industry is generative AI, specifically large language models (LLMs), such as ChatGPT. While generative AI leverages patterns within existing data to fabricate new, unique data sets, LLMs take this concept further, offering unparalleled capabilities to dissect and orchestrate intricate information and produce interactions akin to human dialogue.

The manufacturing sector deals with vast and complex unstructured data, including sensor readings, images, videos and telemetry. Real-time streaming and integration with contextual data sources are crucial for meaningful responses to events.

LLMs can revolutionise the industry by empowering personnel with enhanced tools. They can redefine how operators engage with systems and documents, driving exponential improvements in productivity, customer satisfaction and financial performance.

Two often overlooked areas are essential: natural language interfaces and product design and optimisation. These areas hold immense potential, delivering tangible impact and significant returns on investment in manufacturing.

AI, particularly LLMs and their natural language interfaces, have vast potential to revolutionise manufacturing efficiency, worker engagement, product quality and adoption.

Manufacturing facilities require seamless information transmission, often achieved through production reviews. These reviews aim to uncover discrepancies, enhance decision-making and improve operational efficiency, customer satisfaction and fiscal outcomes. By shifting to targeted human-like dialogues, organizations can focus on identifying bottlenecks, strategising recovery plans and reducing complex data extraction time. This can streamline processes, leading to enhanced operational performance and productivity.

LLMs play a crucial role in this transformation, enabling operators to interact with complex systems, such as digital twins and control towers, using natural language. LLMs improve voice interaction accuracy, making it accessible and repeatable in noisy environments. This enhances productivity by reducing the learning curve and eliminating the need for extensive data analytics or coding training. Non-technical personnel can now navigate intricate systems, leading to improved responsiveness and adoption. LLMs redefine human-machine interactions, offering transformative benefits to the manufacturing industry.

Sophia Velastegui, Chief Product Officer at Aptiv, who has successfully leveraged AI innovation to further several global businesses, shares “LLMs can be integrated into user interfaces to facilitate human-machine interactions. They have the potential to revolutionise this interaction, making it as simple as having a conversation. Furthermore, they could significantly improve safety, as workers can focus more on their tasks and less on deciphering complicated instructions. Businesses benefit by leveraging these subject matter operators for creative solutioning, versus repetitive tasks.”

So, LLMs act as a vital conduit, fostering enhanced collaboration between operators and machinery through natural language interfaces. In doing so, they democratise the accessibility of complex systems, propelling a remarkable surge in efficiency and productivity.

Traditionally, designers focus on product concepts and specifications, while operators handle manufacturing tasks. Leveraging LLMs, however, enables a more informed and democratic design process, incorporating insights from frontline operators. These operators possess practical understanding and can contribute valuable insights. LLMs assist in translating their recommendations into actionable design suggestions.

By analyzing operators’ insights, LLMs generate design alternatives that consider practical considerations and constraints, resulting in realistic and effective solutions. This collaborative approach fosters a sense of ownership and engagement among operators.

LLMs have a significant role in product design and optimisation due to the wealth of information within manufacturing equipment and robot systems. They combine this knowledge with market trends, scientific literature, evolving ESG considerations and customer preferences.

The resulting design concepts align with sustainability and environmental guidelines, suggesting alternatives, simulating performance scenarios, and recommending sustainable materials and manufacturing processes. This integration of data and expertise drives innovation in design while addressing environmental concerns and meeting customer demands.

“This kind of integrated design-manufacturing approach could be a game-changer in the industry,” reflects Dr Rana el Kaliouby, General Partner at AI Operators Fund and ardent advocate of humanising technology. “LLMs – especially multi-modal LLMs that can take text prompts and produce images or design renditions – can accelerate product ideation, leading to more effective, practical and human-centric product designs in the manufacturing industry.”

Collaborative manufacturing with LLMs presents numerous advantages. It is, however, crucial to address intellectual property rights, ownership and trade secrets to protect the interests of all stakeholders involved. Clear guidelines, policies and frameworks should be established to ensure alignment with existing laws and corporate governance.

Cathy Li, Head of AI at the World Economic Forum, emphasises, “While generative AI offers vast opportunities for value creation in industry, there are multifaceted risks and challenges to consider. Clear knowledge about the LLMs’ composition, in terms of safety guardrails, testing and evaluation and risks, as well as data provenance, is key to making sure that it is aligned with existing laws and corporate governance. Also, proper training to end users is key to fostering responsible and informed use of generative AI-powered applications.”

Cynthia Hutchinson, CEO of the US Center for Advanced Manufacturing, highlights the centre’s role in fostering collaboration among industry, government and academia to safeguard interests in AI-enhanced manufacturing. She says: “It is important to create an inclusive ecosystem in advanced manufacturing, ensuring all stakeholders have a voice and establish trust for knowledge exchange, innovation, addressing challenges and driving economic growth. There is tremendous potential for leveraging expertise and resources from each sector to unlock the full potential of AI and LLMs, like ChatGPT, contributing to a more sustainable future.”

Raising awareness and promoting the adoption of AI in the manufacturing industry is critical for the continued growth and success of the manufacturing industry. To fully realize the potential of AI in manufacturing, we need further research, discussions and industry case studies to shed light on untapped applications.