The European Central Bank is exploring the use of ChatGPT-like large-language models for things such as document analysis and software testing.
The ECB is "cautious" about the use of AI and is looking at questions of data privacy, legal constraints and ethical considerations, says chief services officer Myriam Moufakkir in a blog.
However: "By putting in place the appropriate governance, coordination, infrastructure and investment, we will pull together the various strands of our work on AI and accelerate its adoption across our organisation."
Among the possible uses of LLMs the central bank is exploring is the writing of initial drafts of code for experts for use in analysis and the testing of software. The models can also analyse, summarise and compare documents prepared by the banks the ECB supervise and help prepare summaries and draft briefings.
A large language model can also help improve texts, making the ECB’s communication easier to understand for the public. The bank already uses neural network machine translations to help communicate with European citizens in their mother tongues.