Unlocking the power of generative AI

GENERATIVE artificial intelligence (AI) has taken the world by storm, capturing imagination and consuming fears in equal measure. Although generative AI with massive Large Language Models (LLMs) is still in its infancy, the technology is beginning to cement itself in several applications and sectors. In fact, AI applications are estimated to create a future productivity dividend ranging between US$2.6 trillion and US$4.4 trillion annually for the global economy.

For businesses, the benefits of generative AI are immense. The technology has proven itself a powerful tool in cybersecurity, in the automation of routine tasks, and even as a personal assistant in day-to-day office work. It has also led to immediate wins such as helping create synthetic data that simulates more difficult-to-acquire real data.

Yet, it is a different story when it comes to industrial innovation, 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 around the corner that stakeholders might have been led to believe.

Organisations looking to establish generative AI as an integral part of everyday business will need to exercise caution, and address its potential issues and limitations. Only then can they avoid the two dangers of emerging technologies – gold-rush frenzy and development inertia – and embrace the technology fully.

Preparation is power

To effect lasting change, organisations will need to combine LLMs with up-to-date data from additional AI technologies – such as neural networks and monitoring agents – to ensure data dependencies remain current and contextual for maximum, real-world value. This way, organisations can get quick, out-of-the-box generative AI solutions that are fit for purpose with the ability to scale when needed. Only with such a combination of advanced technologies can frontline AI move from task-based to objective-driven applications.

Moreover, it is important that the data is carefully selected, cleaned and refined to improve its reliability and accuracy. The data selection process is crucial as it teaches an algorithm how to make accurate predictions, by identifying relevant and representative datasets for specific purposes. This dataset becomes the benchmark against which data scientists and developers assess the accuracy of a system’s predictions and generated output.

For relevant use-cases where existing datasets are limited, it is worth considering applying data augmentation to increase the diversity and sample size of the applied dataset. This is also applicable to computer-vision based applications in medical devices, and in industrial automation, where image-capturing is a critical component for quality analysis and quality-control processes.

People are power

Over the past 20 years, AI has made significant transformations in the industrial workplace. The challenge now is to bridge the gap between AI technology and natural human understanding. Generative AI can help bridge this gap, but more needs to be done.

Potential limitations such as data security, and hallucinations – where a LLM AI bot generates incorrect information in a confident and convincing manner – need the recognition of natural human intuition.

For organisations to derive maximum value of AI, the human intermediary cannot be removed. There is immense value in investing in internal expertise – training data scientists, software engineers, and domain experts. In addition, organisations can establish new functional roles akin to internal product managers. These are individuals with both domain and data science experience, and who will have direct responsibility for overseeing algorithm development routines.

Getting the most out of Gen AI tools

For industrial businesses already exploring generative AI tools or processes, here are a couple of best practices to aid effective implementations.

First, organisations must engage with the generative AI communities to be kept up to date with the latest directions – be it research papers, conferences, forums, or industry publications.

Next, define relevant use-cases that can shine a light on your organisation’s needs, challenges, and opportunities. A thorough analysis can help determine the areas where generative AI could have the most impact and align with your business and sustainability objectives. This ensures a targeted and customised approach in implementing generative AI tools.

Lastly, organisations should establish an ethical framework to guide the responsible use of generative AI, one that complies with the relevant regulations and standards, and prioritises user privacy and data protection, in the jurisdiction.

We are just getting started with AI, whether generative or other kinds. While the possibilities appear boundless, we must recognise that harnessing its full benefits will require time and effort. As regulatory frameworks mature, we can define the boundaries and create the necessary safety thresholds, to build AI more responsibly.

The exciting path ahead is filled with possibilities. Only by addressing the challenges around generative AI and LLMs can we fully unlock its transformative power.

The writer is global head of AI and advanced analytics at AVEVA