Dear Colleagues,
As mentioned at our last Council meeting in Tirana, EFPA is working actively on the theme of digitalisation. A call has been sent out for nominations from Full Members to the new Ad Hoc Working Group Digitalisation (please respond by Friday June 21): See EMC 8/2024
As another step, a document giving interim guidance on the use of Generative Artificial Intelligence (AI) in the course of EFPA duties has been prepared. Please find it here attached and in the Workplace Knowledge Library for your reference. These guidelines are just a starting point- with the help of the newly formed Digitalisation Group a further review will be carried out of some of the various implications of the internal use of AI and other digital tools, with updated guidance prepared.
For the Board of Directors
https://efpa.workplace.com/work/file_viewer/1112424633154834/?surface=POST_ATTACHMENT
Interim guidance on the use of Generative Artificial Intelligence (AI) in the course of EFPA duties Generative AI offers a range of opportunities, including assisting non-native speakers in producing texts in multiple languages, quickly summarising information from different sources, and effectively managing extensive knowledge repositories. However, it also poses risks with significant consequences, such as the potential spread of misinformation and unethical practices. As the use of AI becomes more prevalent, there are a range of current and future uses. While its application likely presents fewer problems in uses like text rephrasing or translation, concerns emerge in other areas. These include the risk of biases coming from false or context-insensitive information provided by AI, as well as breaching confidentiality by uploading sensitive data or data that constitutes intellectual property into AI systems. The use of certain AI systems and paid subscriptions may provide an element of assurance to avoid some of these risks. Nonetheless, through its use the system may learn from the inputs it has used even if that data is later withdrawn. Paid for systems cannot therefore be regarded as a fail-safe solution to these issues. It is recognised that both the legal frameworks and the development of guidance by many organisations on the use of generative AI are at an early stage, the tools are developing rapidly and much remains to be learned. In this context, EFPA has developed the following interim guiding principles regarding the production of materials using AI: • Transparency and Accountability: openly disclose the use of AI, especially when it impacts decision-making processes, and take responsibility for the materials published. This involves being clear in any EFPA document where AI has been used, and how. • Quality and Integrity: Make sure the data inputted into the AI system is accurate, and understand and explain clearly AI’s limitations regarding the use of the output. Review the output to ensure it makes sense in the light of other sources of information and knowledge, triangulating the AI output with other sources. Make sure both authors and users of a document can trace AIgenerated results back to their source. • Privacy and Security: respect individuals’ privacy, maintain confidentiality and strictly adhere to GDPR regulations. Avoid entering personal or sensitive data into any AI tools and avoid uploading unpublished work, particularly if it contains sensitive information or information that constitutes intellectual property. • Fairness: endeavour to remain informed about any AI application that is used to mitigate biases and ensure fair outcomes. Aim to assess mechanisms implemented for ongoing bias evaluation and correction, while also considering the limitations of generative AI.
