Who’s leading the global AI race?

The data is collected by AI Index using diverse datasets that are referenced in the 2020 AI Index Report chapters.

Methodology

Step 1: Obtain, harmonize, and integrate data on individual attributes across countries and time.
Step 2: Use Min-Max Scalar to normalize each country-year specific indicator between 0-100.
Step 3: Take arithmetic Mean per country-indicator for a given year.
Step 4: Build modular weighted Mean for available pillars and individual indicators.

Aggregate Measure

The AI Vibrancy Composite Index can be expressed in the following equation:

AI Vibrancyc,t = ( Ψpillar × [Ωc,t × Ψindicator ]) ÷ N

where c represents a country and t represents year, Ωc,t is the scaled (0-100) individual indicator, Ψindicator is the weight assigned to individual indicators, Ψpillar is the weight specific to one of the three high-level pillars and N is the number of indicators available for a given country for a specific year.

Normalization

To adjust for differences in units of measurement and ranges of variation, all 22 variables were normalized into the [0, 100] range, with higher scores representing better outcomes. A minimum-maximum normalization method was adopted, given the minimum and maximum values of each variable respectively. Higher values indicate better outcomes. The normalization formula is:

Min—max scalar (MS100) = 100 × (((value) − (min)) ÷ ((max) − (min)))

Coverages and Nuances

A threshold of 73% coverage was chosen to select the final list of countries based on an average of available data between 2015-2020. Russia and South Korea were added manually due to their growing importance in the global AI landscape, even though they did not pass the 73% threshold.

Metric Definitions

  • Research and Development
  • Economy
  • Inclusion*

Author: Carla

Deixe uma resposta

O seu endereço de email não será publicado.