evidoria

← Back to browse

Good practice Imported

Citizen Lottery Jury — South Korea's Public Evaluation Panel for Selecting Its Sovereign AI Foundation Models

South Korea · Seoul · See the South Korea profile · See the Seoul profile

Evidence: Descriptive / self-reported Top 25% 73/100 · Ask Evidence Copilot about this practice

South Korea's MSIT recruited 200 citizens by demographic lottery to test four rival sovereign LLM consortia (Aug 8-11, 2026) as part of a $5.7bn National Growth Fund programme that will pick two national AI champions by December 2026.

200 citizens
Citizen evaluators selected by demographic lottery (Aug 2026)
5.7 USD billion
National Growth Fund allocation for the sovereign AI stack (May 2026)
3.5 USD billion
MSIT 2026 AI budget (2026)

Details

Maturity
Pilot
Promoter
Ministry of Science and ICT (MSIT), Republic of Korea
Period
2025-2026
Keywords
AI policy, public procurement, technology governance

Context

South Korea's Ministry of Science and ICT (MSIT) is running a Sovereign AI Foundation Model project, launched in August 2025, to select two national AI champions from a shrinking field of competing consortia (LG AI Research, SK Telecom, Upstage, Naver Cloud, NC AI, and later Motif Technologies) by December 2026, backed by roughly $5.7 billion from the National Growth Fund and a 2026 MSIT AI budget of about $3.5 billion.

Objectives

To select two sovereign AI foundation models to serve as the technical backbone of Korea's national AI infrastructure, using a citizen lottery panel alongside technical criteria as part of the evaluation process.

Activities

From August 8-11, 2026, MSIT ran a four-day citizen evaluation phase in which 200 citizens selected by demographically-weighted lottery (rather than technical expertise) tested the four remaining consortia's models on real-world usability and agentic tasks, feeding into narrowing the field from four to three ahead of a final selection of two models by December 2026.

Results

As of the source material, the evaluation window had just closed and no results had been published; it remains unproven whether a lay-citizen panel can meaningfully differentiate foundation-model quality at a technical level, and the demographic-weighting methodology behind the lottery has not been made public.

Implementation
Indicative cost
Very high (> €5M)
Time to results
Medium (1–3 years)
Staffing & skills
Ministry of Science and ICT (MSIT) manages the overall selection process, 200 citizens recruited by demographically-weighted lottery serve as lay evaluators (not technical experts)

Conditions for success

  • Demographically-weighted lottery selection intended to give the evaluation broad representativeness
  • Evaluators tested models on real-world usability and agentic tasks over a structured four-day window

Common failure modes

  • Evaluation window had just closed with no results published, so it is unproven whether a lay-citizen panel can meaningfully differentiate foundation-model quality at a technical level
  • The demographic-weighting methodology behind the lottery has not been made public

Where it fits

Governance type
national government ministry
Scale
national
Income level
high-income

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

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