Innovation Insider, let's talk with Julien Molez about scaling up data and AI
Claire Calmejane
C-Level Executive | Driving strategic vision, growth and operational success across EMEA markets | AI | Digital Assets | Technology | Financial Services | Angel Investor
With an entrepreneurial mindset, Julien contributed to the creation of 2 consulting firms, leading them through their scale-up phase. Because of his strong background in IT engineering through his years at IBM and his robust financial service background built thanks to his many assignents in all parts of the bank, it was a no brainer to hire him as our lead in Innovation to scale up Data and AI across the Group. We asked a last generation artificial intelligence to find a weakness in Julien, and it still has not answered. No one knows if it went on loop because of a bug or if it will end up to be 42.
Julien, could you please tell us about the data and AI strategies and the changes under your leadership?
The main change under my leadership is that, with the support of our CEO, we created a Group AI team that completes and augments our existing decentralized AI set-up to bring more consistency to our Data/AI strategies.
In my opinion, AI and Data are not an end in themselves and must be at the service of a global digital business strategy. Their potential must be understood and mastered by our businesses and functions. That’s why I am convinced that AI use cases must be designed and implemented as close as possible to business needs.
But in the meantime, we know that AI use cases are often highly reusable across business lines (conversational agents, chatbots, document analysis, personalisation) and that reuse is a key factor to speed up implementation.
My team was then positioned on three key topics: setting-up a Group strategic dialogue on AI, increasing sharing on AI topics and raising awareness of all employees.
What are the main challenges to scale AI across the Group?
Unfortunately, there are many! Like any organisation with a strong IT legacy, we have issues on data fragmentation, data access and data quality that sometimes make our Machine Learning developments slower. We also must invest in our MLOps (Machine Learning Operations) to fasten and reduce the production costs of our models and ensure they integrate easily with our existing systems to bring the changes to our clients and end users.
But our greatest challenge is on strategic appropriation of AI to make sure that all key business people have a good understanding of the transformative potential of data and AI and how and when it can be used to transform business processes and value proposal to our clients.
What are your main achievements in scaling-up AI/Data to date?
I have worked on three main enablers to ensure the whole organisation is moving forward on AI:
1. Set-up a frequent and high-quality strategic dialogue with the Group general management and the 25 Business Units/Service Units heads to ensure that Data/AI topics are on their agenda. For example, we asked every Group Management Committee’ (CODIR) member to sponsor a use case and to share it with their peers. We also trained them on Machine Learning (ML) fundamentals during reverse mentoring sessions with ML experts.
2. Strongly increase the sharing of our AI assets across the Group. We have created work groups to share ideas and AI assets on critical domains: NLP for operational efficiency and AI for customer experience. Each of them gathers more than 100 business and data experts. We also created a shared platform that list all Group AI use cases and that is accessible to any employee to encourage ideation and reuse.
3. Raise awareness of all employees on data/AI topics by letting them know what is happening in this field in our different businesses but also by giving them the opportunity to upskill themselves, notably through a very impactful partnership with Coursera.
What’s your commitment to building a more sustainable future in AI and Data?
As data and AI are progressively leveraged to augment our digital proposition to our clients and end users, I’m convinced that we have a great responsibility to ensure we make a fair and transparent use of them.
Within Societe Generale, we are working on the definition of a Group framework on key topics (coding & modelling guidelines, explainability framework, ethic principles, data access process) but also on a robust governance to put all developed models under control and to make sure that any use of sensitive data is managed. I contribute to all these topics alongside all Group experts (Data office, risk, compliance, human resources, CSR).
I truly believe that beyond principles, we must work on inclusive AI and ensure diversity is represented at any stage of AI development. We cannot build tomorrow without a diverse talent base being part of its construction. That’s why we partner with institutions such as WiMLDS (Women in Machine Learning and Data Science) and Women Forum’s Daring Circle in favor of female inclusion in AI.
Where do you see our AI transformation in 5 years?
My goal is to make sure that in five years, AI has become a commonly-used technology that is part of our standard digital tools that our businesses and functions know when to use. I imagine it to be part of massively digitized processes to boost their performance and guarantee that we deliver reactive, intuitive and highly personalised services to all our clients.
I’m also convinced that we are going to see new NLP (Natural Language Processing)breakthroughs, in the line of GPT-3 (Generative Pre-trained Transformer 3, an autoregressive language model), that will force us to keep on reinventing the way we work and interact with our clients.
3 key words to describe yourself?
Optimistic, Everyday learner, Data-driven ??
Do you like this article? share and/or comment. And do not hesitate to suscribe to the Innovation Insider newsletter. Let's share trends and insights from leaders, partners and co-workers inventing the future of banking. Not to be missed!
To read more in InnovationInsider interviews, here are the links:
Conseiller indépendant
4 å¹´Claire, having been lucky enough to work with Julien, I can tell you this is not because of a bug.
Apporteur de solution à des dirigeants de PME décidés à booster leur développement commercial.
4 年Hello Claire Calmejane and Julien Molez. Good to see that Société Générale is actively using its data for AI. you do not mention this in the interview one pain point we often hear : that AI teams are limited by Data Protection regulation in the access to the data they would like to use to train their models. If you could unlock higher efficiency and more use cases by taking a new approach to Privacy management, Laurent Maufras du Chatellier and me will be happy to support you !
Driving Financial & Technology Innovation | SimCorp - Deutsche B?rse | AI & Fintech Sales | Passionate Athlete
4 å¹´What a nice interview! I can only agree with Julien. I spend a lot of time educating people about the myriad uses and benefits of AI, NLP, and data. Above all, i think that the combination of structured/unstructured data and AI makes an immense difference for the asset-/portfolio management in risk analysis, providing an complete overview of the markets, is helping with trend recognition and even helps you to determine direct investibles. And there is more ?? Claire Calmejane
Directrice Communication SG SOCIETE GENERALE Ile-de-France Sud
4 年Top la vidéo ??????