AI
Beyond ChatGPT: How purpose-built AI can transform exhibition research
ChatGPT excels at conversation, but when exhibition organisers need to analyse global markets, they quickly hit their limits. Here, Matthias Tesi-Baur explains MBB’ s new AI Data Scanner tool, and how it could transform data collection for the industry.
I is everywhere, writing
A emails, generating reports, and answering almost any question in seconds. Tools like ChatGPT and Perplexity have made searching and summarising information effortless. But when research depends on accuracy, structure, and context, like analysing hundreds of exhibitions across the world, the difference between a general AI tool and a purpose-built system becomes striking.
Years of working with exhibition organisers and venues have shown us how difficult it is to gather reliable data across markets. Collecting information from hundreds of sources is slow, messy and inconsistent. Each website uses different terminology; categories vary by region; and what looks like a simple comparison quickly turns into weeks of manual checking.
That challenge led us to develop the MBB AI Data Scanner, a proprietary system designed to make exhibition research faster, deeper, and more cost efficient. While it includes artificial intelligence, the Scanner is not an“ AI chatbot” nor another version of ChatGPT. Instead, the AI within the Scanner is there to speed up the parts
“ Comparing ChatGPT and the MBB AI Data Scanner is somewhat like comparing a journalist and a researcher” that can be automated without losing quality. The Scanner extracts publicly available exhibition data, cleans and organises it, spots patterns, and flags inconsistencies for our analysts to review. It doesn’ t simply guess, as ChatGPT has been known to do. And that’ s where it differs completely. While ChatGPT and similar tools can handle moderate amounts of information, they struggle with the scale and complexity of big data quantities, thousands of shows, constantly changing websites, and inconsistent structures. It simply can’ t retrieve the data we are analysing.
Data quality is where many AI tools fail. Scraping information from exhibition websites is a task that LLMs( Large Language Models) like ChatGPT struggle with. If ChatGPT cannot find the data you request, it may in fact fill in the blanks itself, offering data that has no basis in reality whatsoever.
While ChatGPT operates on probabilities, the Scanner operates on proof.
To show how different these two worlds really are, we ran a simple experiment, giving ChatGPT and the MBB AI Data Scanner the same research tasks an exhibition analyst might face. Our test industry: aviation.
1. Finding aviation exhibitions and conferences First, both systems were asked to find aviation-related events worldwide.
ChatGPT pulled limited data and produced listings and categories, with overlaps, outdated entries, and mixed event types. In the end, it explained“ why” the numbers didn’ t align instead of resolving the inconsistencies.
The MBB AI Data Scanner approached the same question differently. Using hundreds of public sources, it cleaned, structured and filtered the data, removed duplicates, aligned event names, and classified them consistently. The result: 167 www. exhibitionworld. co. uk Issue 6 2025 11