Adventure Table - GPT Exercise: Practical AI Engagement with ChatGPT

1 December 2023
Authors Lalo Luna

"Everyone, from pioneers to beginners, is in the process of pinpointing gaps, potential risks, and biases."

3 min read
Adventure Table - GPT Exercise: Practical AI Engagement with ChatGPT

The buzz around Generative AI has taken the tech world by storm. Yet, both agencies and clients alike are still in the infancy stages of exploration and comprehension. There's a vast expanse to navigate when it comes to understanding the monumental potential of Generative AI in the research process. 

Many players in the industry, particularly on the agency side, have already taken their initial steps into the world of Generative AI. They've even begun devising solutions to augment their service offerings. But it's undeniable that uncertainties remain. Everyone, from pioneers to beginners, is in the process of pinpointing gaps, potential risks, and biases. 

On the client front, the scene is diverse. Not every company has jumped on the bandwagon with the same enthusiasm. While some are actively adopting tools and collaborating with partners, many remain in the preliminary phase of experimentation. Companies today are in search of secure environments where their teams can harness this technology. The aim? Seamless integration of various data sources and nurturing their 'knowledge gardens'. 

Discussing real-world applications, Large Language Models (LLMs) like ChatGPT find themselves being increasingly used for consultation, optimisation, and data processing. Interestingly, the Insights teams are not always leading this charge. But a word of caution – not every organisation is on board with letting their teams use ChatGPT or other LLMs. Those who do, especially those utilising free versions, tread carefully due to concerns around confidentiality. 

Among those who are diving into this AI exploration, there are two clear camps. One comprises technical experts who are actively experimenting with various LLMs and more experts in data engineering and the technical process. The other consists of novices - like me - still finding their footing. For both, the pivotal challenge lies in crafting the perfect prompt – one that takes into account the biases and risks of the technology while steering the AI towards the desired result. 

This brings us to the crux: prompt engineering. As we look ahead, it's evident that roles within insights will undergo a transformative shift. Those of us who are content and data experts will soon find ourselves transitioning to become prompt engineering connoisseurs. 

During the roundtables, we delved into this fascinating realm. We conducted exercises using rudimentary prompts, igniting sparks of ideation and lateral thinking. We even ventured into creating a product concept from scratch. Simple as these exercises were, they were immensely enlightening and well-received by all. 

Related

6 December 2021 2 min read
21 June 2023 6 min read