Constructing an Artificial Intelligence Index using benefit of the doubt approach

In the study, the Artificial Intelligence Index has been constructed, in order to measure the capability, capacity and potential of countries in the context of uptake, usage, application and development of artificial intelligence technologies. The Artificial Intelligence Index, has been constructed for 120 countries using 16 sub-indicators in 5 categories, enables the relative determination of the current status of the countries, comparison of their positions and identification of country groups with similar artificial intelligence capability, capacity and potential. In the study, in the weighting of the Artificial Intelligence Index sub-indicators, instead of using a specific weighting scheme, Data Envelopment Analysis based Benefit of the Doubt Approach weighting method, which takes into account the policy priorities and inherent capabilities of the countries, has been used. With the Benefit of the Doubt Approach, country-specific optimal weights that maximize the composite index value for each country have been structured based on the observed performances of the countries. In the study, the Artificial Intelligence Index scores have been calculated for 16 alternative weighting scenarios. As a result of the evaluations, the scenario that has higher discriminatory power in revealing the relative artificial intelligence capability, capacity and potential of the countries and reflects the information represented by the 16 Artificial Intelligence Index sub-indicators in a broader scope, has been determined. Uncertainty and sensitivity analysis have been conducted to assess the robustness of the index scores and country rankings obtained with this scenario. As a result of the analysis, it was concluded that this scenario provides a reliable picture of the artificial intelligence capability, capacity and potential of countries at the global level. In the study, using the sub-indicator and category shares obtained with this scenario, the relative importance of sub-indicators and categories on a country basis has been explained and the main drivers, strengths and weaknesses of countries have been determined. In addition, the relationship between the Artificial Intelligence Index (scores and category shares) and economic development indicators has been analyzed in order to investigate the effect of artificial intelligence capability, capacity and potential levels of countries on their economic development.