Maximizing generative AI’s value for manufacturers

By combining LLMs with real-time industrial data and customized AI technologies, it is possible to unlock their true transformative power.

By Jim Chappell February 24, 2024
Image courtesy: Brett Sayles

Generative AI insights

  • Generative AI’s impact on industry extends beyond quick wins, requiring integration with up-to-date industrial data and purpose-driven AI for enduring value, reshaping operations, enhancing efficiency, and promoting sustainability.
  • Challenges with generative AI, including hallucinations, security, cost implications, and biases, necessitate responsible implementation. As AI evolves towards general AI, ethical considerations, regulations, and human supervision become crucial for sustained coexistence and benefits.

In 2023, artificial intelligence (AI) tools really started to grip the corporate world. Generative AI, in particular, appears to be able to meet every possible business need.

Although generative AI has been around for some time, the big change has been in the training of massive large language models (LLMs) and it is these tools that are likely to significantly disrupt most industries.

Generative AI has also proven to be a powerful tool in cybersecurity and in the automation of mundane tasks. It has led to quick wins such as helping create synthetic data that simulates more difficult-to-acquire real data. For training and testing machine learning, experts can now create larger and more diverse datasets.

It’s a different story when it comes to industry – where creating enduring value will require companies to look beyond quick wins and low-hanging fruit. On its own, generative AI cannot spark the kind of revolution stakeholders have been led to believe is around the corner.

To effect lasting change, industry needs to be able to blend LLM “brains” with up-to-date industrial data together with additional, fit-for-purpose AI (such as neural networks and monitoring agents) to redefine a new realm of value specific to the industrial world – all with minimal set-up and configuration.

Only with such a combination of advanced technologies can frontline AI move from task-based to objective-driven applications.

Ultimately, we see the different forms of AI driving industry toward improved industrial efficiency and sustainability. For example, AI could be tasked with minimizing industrial greenhouse gas emissions using legal, ethical, and safe methods. The technology would work to achieve this goal by identifying underperforming assets, coordinating maintenance, and enhancing operational control to optimize fuel utilization and reduce harmful atmospheric emissions.

Of course, humans will need to closely supervise the initial stages before AI can do the heavy lifting and support the move to full autonomy. This is just one example of how AI will play an increasingly critical role across all industries, revolutionizing operations and promoting sustainability.

Four key challenges with generative AI

Responsible implementation remains vital to achieving these goals, and businesses will need to tackle questions around legal and ethical factors, cost issues, and safety and cybersecurity risks. These include:

  • Hallucinations: Where LLMs provide convincing but incorrect responses. For now, these generative pre-trained transformer (GPT) models are improving, and the right prompt can elicit much more useful responses.  In addition, by combining GPT models with up-to-date industrial data, incorrect answers are greatly reduced, with significantly improved results in the industrial context.

  • Security: Confidential data can be compromised when sent to LLMs that are used for training at large. In some ways, these concerns are comparable to initial worries around putting enterprise data in the cloud – this is now common practice governed by adequate security protocols. In addition, vendors such as OpenAI are already working to avoid misuse. Similarly, Microsoft’s new Copilot is architected to protect tenant, group, and individual data, while taking advantage of existing Microsoft security, compliance, and privacy solutions.

  • Cost implications: Generative models are computationally expensive and training larger and more complex models requires significant computing resources. To offset that outlay, it is important to develop models that ensure reliability and consistency, particularly as they can be created more rapidly and begin to merge with self-learning systems.

  • Bias or discriminatory language, misinformation, malicious and/or fake information, and privacy and data security: For now, researchers and policymakers are exploring ways to mitigate these risks, such as by developing methods to detect and remove biases, improving transparency and accountability around the use of generative AI, and implementing appropriate legal and regulatory frameworks. A robust model with clear guidelines around transparency and accountability must be developed and agreed upon to allay fears about content validity or the use of synthetic data.

For those industrial organizations already onboarding generative AI, there are several best practices for implementing it effectively. First, knowledge. We suggest that organizations remain informed and engage with the generative AI community, ensuring they are up to date with the latest developments through research papers, conferences, forums and industry publications.

Likewise, customization is essential. Consider defining relevant use cases that can shine a light on an organization’s needs, challenges, and opportunities to add value. A thorough analysis can help determine the areas where generative AI can have the most impact and align with your business and sustainability objectives.

An increasing number of software vendors are integrating generative AI and LLMs into their offers. In this way, organizations can get quick, out-of-the-box generative AI solutions that are fit for purpose with the ability to scale as needed. The most useful solutions combine LLMs with up-to-date data from additional AI technologies such as neural networks and monitoring agents to ensure data dependencies remain up to date and contextual for maximum, real-world value.

Finally, establish an ethical framework to guide the responsible use of generative AI, comply with the relevant regulations and standards, and prioritize user privacy and data protection.

Rapid evolution with generative AI

Generative AI and allied fields will continue to evolve. Alongside, this will be a growing focus on addressing ethical considerations and efforts to mitigate biases, ensure fairness, and incorporate guidelines into the design, training, and deployment of generative models.

This evolution is likely to occur in stages, with humans remaining central to supervisory and strategic roles even when fully autonomous operations are commonplace.

We are rapidly moving from narrow AI toward general AI, where software will become much more human-like in capability. As this trend continues, AI will become more objective-driven, leveraging everything at its disposal to achieve its goal. We must do all that we can to ensure AI is used for good and to improve humanity, such as being instructed to reduce greenhouse gas emissions at industrial facilities. At the same time, regulations and guardrails will be critical in order to continue to sustain humanity and its coexistence with AI.

Business leaders already face a multitude of challenges that demand attention. Generative AI, when integrated with industrial data and other advanced digital technologies, will help provide the solutions needed. At the same time, it is important to remain alert to potential risks, reevaluate our core business processes and consider how to equip workforces with the skills and capabilities required in this new era.

The exciting path ahead is filled with possibility – but it is only by addressing the challenges around generative AI that it will be possible to unlock its transformative power.

Author Bio: Jim Chappell is global head of AI and advanced analytics at AVEVA.