Interview: AI government advisor Inma Martinez

Ian Taylor speaks to the leading government advisor ahead of her appearance at Abta’s Travel Convention in October

Could you separate the hype around ChatGPT from the reality?
“OpenAI’s ChatGPT and Google’s Bard are large language models (LLMs) trained to understand and generate language in grammatically correct sentences and even literary styles. When prompted to answer questions about subjects, ChatGPT searches in its training database of content. This only reaches to 2021 and it does not comprehend all content, for example, academic research papers.

“The hype has emerged from those who pretended to use ChatGPT for tasks it was never meant to do, such as explaining philosophical or ethical dilemmas or proposing solutions to problems based on its ability to think. LLMs do not think outside of language or their ability to write code or generate images from text. The data they would require to do so, to encompass all knowledge in all disciplines, is impossible to feed into their training programmes. That is why ChatGPT has created ‘aberrations’ in its answers.”

What impact do you foresee on jobs and the future of work?
“Generative AI could be used in customer support, content generation, translation, virtual assistants and contexts of quick human interaction because it generates responses that often feel more human-like than earlier chatbots.

“ChatGPT is accessible to developers to integrate into applications and services, and this has created exponential growth in the marketplace for ChatGPT-based products and services. Predictions point to the complete deletion of jobs in professions such as paralegal work and risk analysis at investment houses or journalism. But this is untrue. LLMs are a tool to relieve the burden of certain tasks, but they require human supervision and adaptation of their output to the workstream of a given role.”

What concerns you about AI?
“In my position as chair of the expert group at the G7/G20 AI initiative, we are immersed in creating policies and business principles that will ensure that AI, when it is commercialised and developed, is safe, lawful and focused on human needs. My mission is to ensure AI becomes a force for good and a key factor in our economic progress and social welfare.”

What are the implications of the vast amounts of data required to train LLMs?
“The most concerning is their energy resource intensity. Training large AI models requires significant computational resources, leading to substantial energy consumption, which is exhausting the grid and will force many governments to prohibit their further development if energy management is not implemented.

“Data quality is the second-most concerning. The quality of data in large datasets can vary significantly, including errors, inaccuracies and ‘noise’. Models can inadvertently learn from these, impacting their reliability.”

What are the data privacy and security implications?
“‘Confabulations’ can occur when the AI system attempts to provide an answer or fill in gaps in its knowledge but lacks access to accurate information or context. It may have seemed a great idea to train LLMs to do this, but it creates misinformation which is one of the crucial issues we are trying to fight.

“AI models do not fact-check information before generating responses. They can generate statements that are inaccurate, misleading or even fictional if they encounter gaps in their knowledge.”

What is your advice to business and governments?
“Policies and regulation don’t fix every issue like a Swiss Army knife. If we adopt principles, this is a more viable way to get all parties to agree on what is acceptable and what is not. Right now, the consensus in most nations, especially in the G7 and G20 and their partners, is that these issues with AI have to be dealt with through international cooperation. If ever there was a time when government had a clear view of the importance of dealing with AI and its negative effects, it is now.”

● Inma Martinez is chair of the expert group and co-chair of the steering committee at the Global Partnership on AI (GPAI) set up by the G7 group of countries and Organisation for Economic Cooperation and Development.

She is due to address the impact of generative AI on the travel industry and travel workforce in a keynote presentation to the Travel Convention in Bodrum, Turkey, from October 30 to November 1

Full details and registration:

Share article

View Comments

Jacobs Media Group is honoured to be the recipient of the 2020 Queen's Award for Enterprise.

The highest official awards for UK businesses since being established by royal warrant in 1965. Read more.