The Ministry of Science and ICT on the 27th released the detailed scores of the second-phase evaluation for the independent artificial intelligence (AI) foundation model (Dokpamo) project.
The disclosure was made at the request of Motif Technologies (Motif), a team eliminated in the second-phase evaluation, and with the consent of the other elite teams. Earlier, Motif filed an objection with the Ministry of Science and ICT over the second-phase selection results for Dokpamo, seeking disclosure of the detailed evaluation criteria and a reconsideration.
The SK Telecom elite team topped the second-phase aggregate evaluation with 70.6 points. It was followed by Upstage with 69.9 points, LG AI Research Institute with 69.0 points, and Motif with 65.8 points.
The evaluation was conducted on a 100-point scale, with 40 points for benchmark evaluation (AAII 25 points and NIA 15 points maximum), 35 points maximum for expert evaluation, and 25 points for user evaluation (experts 15 points and the general public 10 points maximum).
By category, Motif ranked first in the Artificial Analysis Intelligence Index (AAII) evaluation, which adopts global standards, with 11.9 points. It was followed by Upstage with 9.4 points, SK Telecom with 8.8 points, and LG AI Research Institute with 7.8 points.
SK Telecom (13.4 points) and Upstage (13.3 points) took the top spots by a narrow margin in the National Information Society Agency (NIA) benchmark evaluation, which covered seven areas including Korean language and reliability. LG AI Research Institute scored 12.8 points, and Motif scored 12.7 points.
LG AI Research Institute received the highest score, 29.5 points, in the expert evaluation, which included 10 external experts from industry, academia, and research. The rankings were SKT (29.3 points), Upstage (29.1 points), and Motif (27.1 points).
SK Telecom (11.6 points) ranked first in the professional user evaluation, part of the user evaluation measuring real-world usability, which included 49 AI startup CEOs. LG AI Research Institute (7.6 points) ranked first in the citizen panel evaluation, which included 185 members of the general public.
A Ministry of Science and ICT official said, "Dokpamo is not about simple ranking competition or selecting teams for elimination, but about the qualitative growth and expansion of the domestic AI ecosystem," and added, "Considering the rapid evolution of AI technology, we are changing development goals every six months (moving target), and the evaluation criteria and methods were also finalized and carried out without disagreement after prior consultations with the elite teams."