From curiosity to collaboration: A data scientist's journey through the JournalismAI Fellowship

Edited by Lakshmi Sivadas

Juan Ginzo, a 2023 JournalismAI Fellowship alumnus, and an AI & Data Science Manager at News UK reflects on his experience with the JournalismAI Fellowship and discusses the transformative role of GenerativeAI in newsrooms.

Juan Ginzo, an AI & Data Science Manager at News UK, started the JournalismAI Fellowship with a mix of excitement and curiosity, eager to explore how GenAI could transform the way journalists work with unstructured data. 

The JournalismAI Fellowship focuses on collaborative projects. It brings together journalists and technologists from all parts of the world to use AI to create solutions that enhance journalism.

Ginzo, who was relatively new to the journalism industry, brought significant technical experience. He applied to the Fellowship alongside George Lindsay-Watson, the Deputy Engagement and Innovation Editor from News UK. They had worked together on previous projects. With Lindsay-Watson's deep experience as a journalist and Ginzo's technical expertise, they were confident in their combined strengths.

Ginzo and Lindsay-Watson partnered with Clare Spencer and Maryam Ahmed - an editorial and technical pair from the BBC who had also applied to the Fellowship. 

Their team aimed to extract valuable insights from the Register of (UK) Members of Parliament's financial interests. They wanted to explore how this technology could extract and structure information in a way that would be useful to journalists to uncover potential stories. 

Picture Credits: Yasmine Boudiaf & LOTI / Better Images of AI / Data Processing / CC-BY 4.0

Setting the stage for success

The team's mindset evolved during the Fellowship. At first, they were optimistic and had broad ambitions, but they quickly realised that focusing on a tighter scope was essential due to time constraints and other professional commitments. 

“Originally we were looking to use both the register and the parliamentary debate transcripts. Later, we decided that if we were to work on this for six months, we had to focus only on the register,” Ginzo said.

GenAI was still new in 2023, and he says it was a team effort to decode it. Particularly, he credited his technical counterpart, Maryam Ahmed, and their assigned team mentor, Ines Montani from Explosion AI.

“The technical mentor gave us ideas on how to do the nitty-gritty of implementing a specific type of AI on a (specific) type of data,” he explained. 

The team’s project eventually reached a satisfactory level of completion, even if it's currently dormant due to time constraints. He said that this tool could be useful for similar applications or problems in the future. Read more about their project here. 

"We got to a good place with the project, we’re able to present and were happy with the results,” Ginzo said. 

Impact

Ginzo and others in the Fellowship have the chance to interact not only within the teams they’ve formed but with the cohort as well. Within teams, primarily the interactions are between editorial and technical members. Ginzo credits this collaborative structure for helping him gain in-depth knowledge of GenAI as well as learning non-technical aspects.

For instance, Ginzo and his team met with the people maintaining the data set at Hansard to understand what they could and could not do with the data. 

“We got to understand GenAI in-depth and a wider variety of tools than I would have without the Fellowship. On the non-technical side, it was really useful to understand how government data works and that's also something I can take forward into other roles.”

He felt that the Fellowship had a long-term impact rather than a short-term one. 

For example, it allowed him to explore a variety of tools. One of his key takeaways was learning about spaCy LLM, a tool that offers additional functionality beyond standard GenAI APIs such as ChatGPT. He explained that the tool's capabilities to handle various NLP (Natural Language Processing) tasks, opened new doors for him and his team. 

Lastly, he emphasised the value of community and collaboration.

"Getting to know different people with varied ideas and experiences, but all working in related areas, was the part that added the biggest value to me," he said. 

It’s been a few months since the Fellowship, and for now, Ginzo is working on building predictive models for article performance and exploring use cases for GenAI in the newsroom at News UK.