• FR

menu
Bannière Hero

How to justify an ethical data approach and above all, one that does not leave political decisions to AI, no matter how intelligent it may be.

Image  MIT validates the SixFoisSept approach

MIT validates the SixFoisSept approach

well, almost 😉

We found this article from MIT very interesting from two points of view: tax-setting policies that are Data-driven (not politically-led) & the need for Artificial Intelligence algorithms to be transparent.

Impôts DataDriven ou décision politique?

Taxation policy: Data-driven or political choice?

Although the percentage of your income that goes towards the taxman may be Data-driven, we feel it is crucial that decisions relating to taxation should be taken by politicians. In this article, the goal was to try to work out the right level of taxation in order to minimize income inequality – which is in itself an eminently political proposition.

 

Algorithmic transparency

When we first founded SixFoisSept, ethics and transparency were very much a part of our vision statement. As a result, no matter how complex the models used in our work are, we must be able to explain them. We need to be able to ensure data traceability, explain why we have chosen the rules that we have etc. In short, we’re convinced that our work needs to be as far removed as possible from the all too real Black Box approach.

SixFoisSept takes a White Box approach!

As the MIT article concludes, if you’re going to use an algorithm to calculate how much tax people should pay, you’d better be able to explain how this algorithm works!

Article du MIT : An AI can simulate an economy millions of times to create fairer tax policy

https://www.technologyreview.com/2020/05/05/1001142/ai-reinforcement-learning-simulate-economy-fairer-tax-policy-income-inequality-recession-pandemic