Despite their impressive capabilities, generative AI models such as ChatGPT are showing gaps in specific industry knowledge. Companies from a variety of sectors, including sports, farming, and finance, are discovering that these AI systems—trained on broad datasets sourced from the internet—lack the depth needed to accurately address their unique needs.
Earlier this year, the PGA Tour experienced this limitation firsthand. During a test, ChatGPT struggled to answer a basic question about Tiger Woods, one of golf’s most iconic figures: how many times he had won the PGA Tour. Scott Gutterman, Senior Vice President of Digital and Broadcast Technology at the PGA, was quick to realise that while AI can access vast amounts of general data, it still fails in delivering deep, accurate insights for specific industries like golf.
“There’s missing data, there’s generalised data. These things have just led to generic responses,” Gutterman said. His experience is shared by many industries relying on AI tools.
As more companies begin integrating AI into their operations, they are quickly realising that these general-purpose models often lack the depth required for sector-specific knowledge. Like a new employee fresh out of orientation, these models require additional training to be effective. Off-the-shelf AI solutions, such as those from OpenAI and Anthropic, are useful but often require extensive augmentation to be truly valuable in specialised fields.
This customisation, however, comes with its own challenges. According to Ritu Jyoti, General Manager and Global AI Lead at International Data Corp, enhancing AI models to be more industry-specific involves increased costs and complexity. Additionally, companies must ensure they have a flawless grip on their internal data, which is often easier said than done.
A critical question for businesses adopting AI is determining how much customisation is required to make the technology reliable enough for their specific needs. There’s a growing ecosystem of support, with consultants, cloud providers like Amazon Web Services, and AI developers offering their expertise to help businesses navigate this issue. However, the journey towards more industry-specific AI solutions is still a complex and costly one for many.
In a rapidly advancing digital landscape, the solution may lie in the hybrid approach that many companies are adopting. By combining general AI models with their own specialised datasets, businesses are striving to create more accurate and useful AI systems, ensuring that they don’t fall short when it comes to detailed knowledge of their sectors.
