New Overview of Artificial Intelligence Arbitrator Systems: Not Appeal, Verification

The rapid increase of scientific publications, especially in the field of machine learning, an important burden of the arbitrator evaluation process. The increasing number of articles led arbitrator quality to drop and increase the workload of observers. big language models (LLM), as a solution tool for this problem, if it is recommended as automatic arbitrator assistants, the success of these systems is measured by incorrect methods.

The new research reveals that LLM-based arbitrator systems should be made radical change in evaluation:

• The available evaluation methods focus on how successful LLMs mimic the articles of human arbitrators; but this does not reflect the real purpose of the arbitrator system
• The new frame offers a perspective of verification-based (verification-centric); this is the forefront of testing scientific validity and nature of an article
• The purpose of the research is to examine the basic functions of the arbitrator process of LLMs—hearsal consistency control, methodology solidity assessment and detecting the original contribution—it can support the actual

The use of LLMs as arbitrator aid is not limited only to provide faster feedback. These systems can improve the quality of the arbitration process and ensure that human observers gain time to make deeper analysis. However, it is necessary to redesign evaluation criteria from the foundation.

This research is an important step on how artificial intelligence can play more effective role in the scientific publishing ecosystem.