Tag: machine learning

In the  fifth and final episode of this five-part podcast series, we’re joined by Dr. Maria Aretoulaki, Conversational and Generative AI Design Lead at GlobalLogic and Patrick Callaghan, Principal Data Strategy Technologist at AWS Ireland.

In this podcast, Maria emphasises the continued importance of design in the context of large language models (LLMs), despite the advances in technology. She discusses the challenges of the unpredictable and sometimes inaccurate outputs of LLMs, emphasising the need for design-led approaches to ensure control, explainability and transparency.

Patrick highlights the crucial role of due diligence in handling data and avoiding biases, emphasising the importance of correct data for successful generative AI outcomes. 

They both discuss the complexity of implementing generative AI technologies on an enterprise scale, emphasising the need for a team of trusted advisors and specialists. The discussion touches on data curation, language engineering, conversation design, and other aspects of the design process. 

Overall, they caution against underestimating the complexity of generative AI and advocate for careful consideration and control over data and outputs to maintain trust and consistency in customer experience.

In the final part of the discussion, previous guests Fawad Qureshi and Tim Marston join Patrick and Maria to offer advice to organisations venturing into Generative AI technologies. Emphasising the need to act promptly, they highlight the importance of carefully selecting use cases that align with organisational goals and future value. 

Stressing that no technology, including large language models, works perfectly off the shelf, they advocate for customisation with human involvement in design and data curation. The key challenge is finding a balance between the opportunities and risks associated with increased data access. 

They suggest organisations should not be paralysed by the fear of misuse but should be mindful and develop tools to address risks without hindering innovation. They caution against adopting technology without a clear business goal or a data culture, likening it to buying a treadmill without a fitness plan – likely to end up unused. 

Lastly, they highlight that a modern data strategy involves more than technology, encompassing a data-driven culture, mindset and organisational structure for efficient data sharing and collaboration.