evidoria

← Back to browse

Good practice Imported

Preserving Icelandic Through GPT-4 — Iceland's Presidential AI Language Partnership

Iceland · Reykjavík · See the Iceland profile · See the Reykjavík profile

Top 66% 53/100 · Ask Evidence Copilot about this practice

On President Jóhannesson's initiative, Iceland partnered with OpenAI so GPT-4 could learn proper Icelandic. Reykjavík firm Miðeind used 40 volunteers to fine-tune the model via human feedback, aiming to keep the language viable in AI-driven public services.

Details

Promoter
Miðeind / Government of Iceland
Period
2022-2023
Keywords
language technology, digital government, cultural policy

Description

Iceland, a small Nordic nation of roughly 370,000 speakers, risks seeing its language marginalised as digital public services increasingly rely on large language models trained mostly in English.
During a spring 2022 visit to the United States, President Guðni Th. Jóhannesson met OpenAI's Sam Altman, launching a collaboration to improve GPT-4's Icelandic performance. Vilhjálmur Þorsteinsson, head of language-technology firm Miðeind, assembled a team of 40 volunteers who applied Reinforcement Learning from Human Feedback (RLHF): for each prompt the model generated four candidate answers, and volunteers selected and refined the best one. The work was announced in March 2023.
The resulting model now underpins Embla, Miðeind's Icelandic voice assistant, and the project was explicitly framed as a template for other low-resource languages. However, no source publishes independent measures of accuracy gains, Embla adoption, or citizen impact, and the effort relied on unpaid volunteer labour and a single private partner rather than a standing government programme.

Read the full analysis: https://www.government.is/diplomatic-missions/embassy-article/2023/03/14/Head-start-for-Icelandic/

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful