‘National AI’ 2nd evaluation results revealed: LG leads expert and public scores

LG AI Research has emerged as the top performer in expert and general public evaluations for the second phase of South Korea's 'Foundation Model' (Dokpamo) AI project. While Motif Technology led in global benchmarks, LG AI Research secured the highest scores from both expert panels and the public.

The Ministry of Science and ICT released detailed results from the second phase of the Dokpamo project, following initial controversy over not disclosing company-specific rankings. In global benchmark evaluations, Motif Technology scored 11.9 points on the 'Artificial Intelligence Index' (AAII) operated by Artificial Analysis (AA), placing it first among the four elite teams. SK Telecom also showed strength, achieving the highest score of 13.4 points in the National Information Society Agency's (NIA) benchmark evaluation.

Expert evaluations, which carried a significant weight of 35 points, saw LG AI Research leading with 29.5 points. This assessment was conducted over approximately five days by a committee comprising external experts from industry, academia, and research, involving written evaluations and Q&A sessions.

“

The Dokpamo project is not intended for simple ranking competition or selecting eliminated teams. Its purpose is to foster the growth of our AI companies beyond competition and to drive the qualitative growth and expansion of the domestic AI ecosystem.

User evaluations showed divided opinions between experts and the general public. SK Telecom recorded the highest score of 11.6 points in the AI expert user evaluation (15 points), while LG AI Research received the highest score of 7.6 points in the general public evaluation (10 points).

The Ministry of Science and ICT stated that the Dokpamo project's objective extends beyond simple ranking or elimination. Its core purpose is to foster the growth of domestic AI companies and drive the qualitative expansion of the national AI ecosystem. The ministry explained that the results were disclosed with a balance between minimizing potential harm to participating companies and upholding fairness and transparency in the evaluation process.

“

We considered fairness and transparency in the evaluation process while minimizing any unexpected direct or indirect damage to some companies due to the disclosure of evaluation results, at a level similar to the disclosure of the first phase evaluation results.