Responsible and Trustworthy AI

Overview of the initiative

Labelia Labs initiated in 2019 a collaborative initiative on Responsible and Trustworthy AI, now regrouping 750+ subscribers to the dedicated Meetup.

This initiative currently spans 3 methodological tools for practitioners:

• The reference framework, initiated in 2019

• The assessment and scoring platform, launched in 2020

• The certification, launched in 2021

To learn more about the initiative, we recommend: our blog article, the podcast (FR) we did with friends at C-Ways, our conference at the B-Boost 2021 (FR) in la Rochelle.

The initiative received financial support from French region Nouvele-Aquitaine, French region Ile-de-France, and the French Public Investment Bank. It is supported by Dataforgood, Aivancity, Blent, SoScience and Impact AI.

Context & History of the initiative

There is a growing tension between the interest in AI techniques and the fears they generate

Interest in AI, data science in general, has been growing for several years. Today this surge has not yet fully materialized in the computer systems in production. But it is getting closer and the acceleration is noticeable.

But where AI seems promising and generates a growing interest in many use cases, it also raises fears and every month that passes sees its share of failed launches and scandals of varying degrees of magnitude. The list of "Awful AI" maintained by David Dao is full of them.

In this context, it is becoming more and more difficult for an organization to implement data science approaches in its products and services and to assume it publicly... To resolve this tension, a framework must emerge to respond to it.

We develop in a participative and iterative way a responsible and trustworthy "data science" assessment

It is by asking ourselves these questions and working on them that we have, from the beginning of 2019, begun to imagine the interest that there would be in exploring this theme and developing a tool that is intended for practitioners, useful and actionable as soon as possible. These preliminary exchanges have turned into a real project, which has become obvious given the Substra Foundation's “raison d'être”: to develop collaborative, responsible and trustworthy data science. 

This is how we launched a participative and iterative approach.

Overview of season 1 (2019 - 2020)

More than a year later, punctuated by five participative workshops and countless exchanges, tests with partner/interested companies, presentations, the undefined object "useful and actionable framework" has taken shape. It became :

• an assessment of responsible and trustworthy approaches to data science;

• for organizations, to assess their maturity;

• composed of about thirty assessment points grouped into 5 thematic sections;

• which provides an output synthetic score on 100 points (theoretical maximum), knowing that as of today 50/100 is a very advanced level of maturity;

• which is complemented by technical resources for each evaluation point, constituting good entry points for organizations wishing to train in it.

Overview of season 2 (2021)

Next step: an assessment platform!

Based on this repository, we designed and developed an assessment platform. 

The primary objective, still with this guiding idea to quickly propose a useful and actionable tool, is to allow interested organizations to do internal work on the theme of Responsible and Trusted Data Science. As we progress through the evaluation and its various sections, we inevitably come up against a subject that we have not yet worked on, which it would be interesting to clear up in order to gain maturity, etc.

You can now freely assess your organization's maturity in responsible and trusted data science!

Plans for 2023!

After three great seasons, Labelia Labs is launching a new cycle of workshops in 2023 with the ambition of continuing to share best practices and feedback, identify new risks and topics to address for a responsible and trustworthy AI!

For this new cycle, Data For Good will continue to support us in the organization of these workshops!

Throughout the year, together, we will offer you different presentations, exchanges, workshops, and roundtables around responsible data science. Don't hesitate to suggest themes or speakers you would like to see!

The workshops will take place on the following dates:

- Monday March 6, 2023
- Monday July 3, 2023
- Monday September 18, 2023
- Monday November 20, 2023