The AI we use, incorporating a Support Vector Machine and Neural Network, determines:
1. whether a task will be automated or augmented
2. by which specific technology in our technology taxonomy
3. when this will happen, based on the adoption curves
Faethm has developed adoption curves that are derived from proprietary modelling.
We model a set of around 150 indicators of technology adoption rates to rank country and industry adoption rates. The ranking is used to reduce the impact of each technology-task adoption curve based on the country and industry that a job is operating in. Our modelling includes a PESTLE analysis per country, per industry, to consider external barriers to implementation as well as other assessments and measures of adoption including historic analysis, survey data, and proxies, to solve for:
a) when the tech will be available
b) the extent of total adoption
c) the speed of adoption
Our machine learning algorithms are highly accurate with both precision and recall scoring in the range of 90 to 99%. This statistic tells us that our modelling is a robust approach to projecting the future impacts of technology.
There are a few factors that are excluded in our technology growth and adoption trend modelling, such as disruptive influences that are difficult to predict, such as war, pestilence and economic recessions.