In a recent episode of DeepTalk hosted by Rémi Tibi, David Giblas, Deputy CEO of Malakoff Humanis and President of Positive AI, shared his vision of how organizations can deploy artificial intelligence responsibly while keeping people at the center of decision-making.

While much of the public debate focuses on technological breakthroughs, model performance and the race for innovation, David Giblas brings the discussion back to a more fundamental question:

What kind of relationship do we want to build between AI and people?

 

Drawing on nearly seven years of transformation at Malakoff Humanis, he explains why responsible AI is not a separate topic from innovation, but a condition for making innovation sustainable, trusted and useful.

From Data to AI: Building Capabilities Before Building Use Cases

 

For Malakoff Humanis, AI did not begin with Generative AI.

The journey started much earlier, with a conviction that data would become a strategic asset for the organization.

Today, Malakoff Humanis protects nearly 400,000 companies and 9 million people. Every year, the Group processes several tens of billions of transactions. At this scale, data is not simply a technical resource. It is part of the organization’s ability to operate efficiently, anticipate needs, improve services and support decision-making.

Recognizing this reality early on, Malakoff Humanis launched a major data transformation around seven years ago.

Rather than treating AI as an external capability, the Group chose to develop internal expertise and retain strong ownership of its data and algorithms. This approach led to the creation of in-house teams dedicated to data science and artificial intelligence, and to the development of around thirty predictive AI and machine learning algorithms designed internally.

For David Giblas, this investment was about more than technology. It was about creating the conditions necessary to understand how AI works, how decisions are made, and how value can be generated responsibly over time.

The lesson is simple: organizations cannot build meaningful AI strategies without first building strong data foundations.

 

Human First: Why Trust Matters More Than Technology

One of the most compelling moments of the conversation comes when David Giblas reflects on the questions that emerged during the early stages of the transformation.

As AI capabilities expanded, governance bodies and employee representatives began asking difficult but necessary questions.

What would these technologies be used for? How would they affect employees? What would remain uniquely human?

These discussions played a decisive role in shaping the Group’s approach to AI.

Rather than focusing exclusively on what technology could do, Malakoff Humanis also chose to define what it did not want AI to do.

This reflection gradually led to a principle that continues to guide AI deployments today: Human First.

The objective is not to automate human relationships out of existence. The objective is to support employees so they can focus on what matters most. This philosophy is particularly visible in customer service.

AI helps advisors access information more quickly, retrieve customer histories, generate summaries and navigate increasingly complex environments. By reducing administrative tasks and simplifying access to knowledge, these tools allow employees to spend more time listening, understanding and supporting customers.

At the same time, Malakoff Humanis has deliberately chosen not to replace customer interactions with generative AI voicebots or chatbots.

For David Giblas, technology should contribute to better service, greater clarity and stronger relationships. Human interaction remains an essential component of trust.

 

Positive AI: Creating the Conditions for Responsible AI at Scale

According to David Giblas, Positive AI emerged from the realization that technological progress alone would not be enough to ensure successful AI adoption.

Organizations also need trust. They need shared principles. They need a framework capable of helping leaders make informed decisions about how AI is developed, deployed and governed.

This conviction ultimately contributed to the creation of Positive AI. Today, the association promotes a practical approach to responsible AI, helping organizations think through the implications of AI before deploying it at scal

One of the key challenges highlighted during the podcast is the transition from experimentation to real adoption.

Many organizations have tested AI. Far fewer have successfully integrated it into everyday operations in a way that creates lasting value.

Scaling AI requires more than technology. It requires clarity of purpose, organizational alignment and confidence from employees, customers and stakeholders.

The DeepTalk conversation offers a concrete illustration of what this looks like in practice: building strong data capabilities, placing humans at the center of decisions, and creating the conditions for trust before scaling innovation.

As artificial intelligence continues to reshape organizations across sectors, these principles may ultimately prove to be among the most important drivers of long-term success.

Because the real challenge is to deploy AI in a way that people are willing to trust.