For a moment, the arrival of generative AI sparked a wave of doubt. If machines can produce images, interfaces, objects… then what is the purpose of design anymore?

In reality, this reading misses the core essence. AIs generate forms, content, and variations at lightning speed. But they do not yet design the relationships that give meaning to these forms: the usages, the contexts, trust, and the compromises between stakeholders.

This is precisely where the role of design becomes more relevant, more demanding… and far more strategic!

1. We are no longer designing objects, but behaviors

Design first focused on form and function, then on usage and meaning. With AI systems, it must now integrate a fourth dimension: behavior over time.

Today, products are becoming learning systems. They observe. They adjust. They anticipate. This aligns with Simon's idea that artificial systems are not merely objects to be manufactured, but artifacts to be designed within a framework of complexity and decision-making.

The smart object can take several forms, and not all share the same level of sophistication. It is useful to distinguish a simple typology:

• Connected Object (IoT): It is linked to a network, reports data, and receives remote commands. However, its behavior remains essentially fixed. We have mostly just moved the interface from a physical button to an app.

• Smart Object: It reacts to conditions with pre-programmed scenarios. For example: "if presence + night → turn on", "if temperature < 19°C → heat". It appears smarter, but it is simply applying rules defined in advance by humans.

• Learning Object: It goes beyond applying generic rules. It learns from your actual usage, detects patterns, adjusts its settings, and proposes automations tailored to your specific context. We no longer program every single behavior; we design the framework within which it can evolve.

• Intelligent Object: It does not just optimize a signal and an action. It considers multiple simultaneous dimensions, is capable of explaining its choices, integrates your feedback, and coexists with other intelligent systems without causing chaos.

It is at this third level that design truly changes its nature. We are no longer content with simply designing a form, defining a function, or configuring a few rules. We design the framework for an evolving behavior over time within a broader social and technical environment. This also echoes Suchman's critiques of the plan model: real-world action always depends on context and situational adjustments, not on purely linear execution.

The designer's role then becomes to:

• define what the object is permitted to learn;

• decide the boundaries of its anticipation and decision-making;

• design how it requests confirmation;

• design how it explains its decisions and accepts correction.

In other words, we are no longer just designing objects; we are designing behaviors over time and, increasingly, the relationships between these behaviors and the humans who live alongside them.

2. Design becomes the design of relationships

A learning object is no longer a static product. It is an evolving system embedded in a network of signs, usages, and dynamics.

Baudrillard previously demonstrated that objects are never isolated. They exist within a system of objects that structures our lifestyles, our status, and our imaginations. Today, with AI, this system does not just signify something to us: it reacts, negotiates, optimizes, suggests, and sometimes insists.

Value no longer resides solely in what an object does, but in how it does it, understands, adjusts, and coexists with us. This is why a relationship-centered approach becomes central. Indeed, Latour insists that the social is not a monolithic block, but an assemblage of associations, mediations, and translations between heterogeneous entities.

The human layer

Between the technical infrastructure—models, data, and protocols—and the visible interfaces, there lies a layer where the quality of the relationship is decided. This is where we determine how the system truly integrates human judgment, oversight, and the freedom to say no.

From this perspective, the interface is no longer just a screen or a button. It becomes a space for governing autonomy. Shneiderman champions precisely this: a human-centered AI where the goal is not to replace people, but to empower them with more control, clarity, and responsibility.

Within this human layer, design must:

• make AI decisions negotiable;

• orchestrate human feedback loops;

• clarify ownership: which decisions belong to the machine, which belong to the human, and which are shared.

We are moving beyond the screen-level UX into the design of an entire relational architecture. We must define who speaks to whom, when, how, through which channel, on what topics, and with what level of trust and reversibility.

3. From interface to alliance

In a world where we must trust the objects around us, humans have every interest in forming powerful alliances with them.

Design then transforms into:

• the design of mutual understanding;

• the design of mental models;

• the design of trust.

The first level ensures the system understands the human better, and the human understands what the system is actually doing. The second level moves away from the fantasy of the "magic black box" to present a clear, honest picture of what the AI knows, can do, and its blind spots. The third level ensures the relationship holds up over time, through errors, updates, and shifting contexts.

In this framework, human-centered AI is not just a moral principle—it is a design absolute. Shneiderman highlights the necessity of reliable, safe, and trustworthy systems that amplify human capabilities without stripping away human agency.

Where Baudrillard described a system of objects quietly shaping us, we are now entering a system of intelligences where something far more delicate is at play: our ability to remain active subjects amidst agents capable of acting, calculating, and learning without us.

Design is no longer just about shaping this system. It is about connecting it, serving as the ultimate champion for keeping humans at the center.

Put simply, we are shifting today from designing interfaces to designing alliances.

4. A simple example: A learning light switch

Let's take a wonderfully mundane object—a light switch—and explore what it becomes when infused with intelligence.

Today, it executes a simple command. Tomorrow, it can learn your routines:

• at what times you turn on the lights;

• under what conditions;

• for which activities.

On a basic level, it can then:

• suggest lighting scenes tailored to your habits;

• anticipate certain needs;

• coordinate seamlessly with other systems.

But moving further, it can also:

• nudge you towards energy savings;

• gently shape your behaviors for the better—but "better" according to whom?;

• introduce you to better options, progressively guiding you toward a new way of experiencing light.

It is no longer just up to you to understand the machine. It is up to the machine to understand how you live in your light.

Remember the days of frantically pressing your TV remote? You thought it would obey better if you pushed harder. In reality, it just needed new batteries. Pressing harder did nothing.

Tomorrow will be different. The switch will understand you. It might turn the light off when you wanted it to stay on. By hitting it faster, harder, or multiple times, you send a clear signal of intent and attention. It will recognize this. But it might also still decide to turn off for your own good.

Design is no longer about placing buttons or functions to create an experience; it's about designing a learning relationship, a truly relational experience (and that is the beauty of service design):

• how far the switch anticipates;

• who defines what is "good";

• when it must ask for confirmation;

• how it explains its decisions;

• how you can correct it and regain control;

• how autonomous it is, and within what boundary of action.

This small example contains everything: behaviors, transparency, learning, and trust.

You are no longer just operating a tool; you are cultivating a relationship. A relationship built on reciprocity, memory, trust, and even disagreement.

With these intelligent objects, you establish a connection that unfolds over time. When you buy a house in 2058, its objects will carry the history of the people who came before you—their habits, their biases, and their unique way of living in the light.

Exciting new types of friction will emerge between the habits of former and current residents, beautifully bringing back to life Baudrillard's concept of the emotional heritage of objects.

5. Typology of uses

This reflection also helps us identify several families of products based on the level of autonomy and negotiation they involve.

• Connected Object: Links, measures, and controls remotely. Focuses on designing seamless access, readability, and continuity between the object and the application.

• Rule-driven Object: Reacts to pre-defined scenarios. Focuses on defining rules, their boundaries, and override scenarios.

• Learning Object: Adjusts behavior based on usage. Focuses on guarding the learning process, ensuring reversibility, and making behavior changes visible.

• Intelligent Object: Evaluates multiple conflicting goals and can explain its choices. Focuses on designing negotiation, trust, the right to contest, and harmonious coexistence with other systems.

This typology keeps us from oversimplifying. A product that "uses AI" might actually be just a slightly more flexible system of rules, or, on the contrary, a truly adaptive system. The role of design changes dramatically based on these levels of autonomy.

6. Scaling up: From object to ecosystem

The light switch is just the entry point. When we scale this up, the exact same logic plays out on a grander, richer scale.

Take a smart factory or office building: robots, heating, ventilation, blinds, lighting, and access control no longer operate in silos. Connected through a shared intelligence, they learn occupational patterns, comfort preferences, and cost constraints together.

Here, design is no longer about drawing a thermostat interface; it is about orchestrating an ongoing negotiation between conflicting interests: energy savings vs. thermal comfort, the morning shift vs. the evening shift, the workshops vs. the corporate offices.

We see the same shift in software. In a SaaS tool, an AI can identify the most frequently used features, simplify journeys based on user profiles, and suggest automated workflows by observing repetitive tasks.

The risk is well known: a system that reconfigures itself without warning destabilizes its users, disrupts their habits, and ultimately erodes trust.

The answer isn't purely technical—it is deeply relational. It comes down to key design decisions:

• distinguishing what can adapt autonomously from what must remain explicit;

• ensuring the system clearly explains what it is changing;

• guaranteeing that users can always hit undo.

In both cases, the individual object dissolves into the ecosystem, yet the core challenge remains unchanged: the quality of the relationship over time.

7. The corporate impact: What this changes for organizations

This shift is not just a design trend; it fundamentally reshapes business models and organizational structures.

First, the product becomes an ongoing service. A learning system does not stop at delivery. It updates, evolves, and reconfigures itself based on data, usage, and regulations. We are moving from a launch-and-leave project mindset to hosting an ongoing, long-term conversation.

Market differentiation is also moving to a behavioral level. It is no longer a list of features that sets two products apart, but the caliber of their interaction over time.

Value is now systemic. Product, data, design, and engineering can no longer be built in silos. A brilliant model without great design leads to confusion; beautiful design without relevant data leads to an empty experience; and data architecture without a user-centric vision delivers a system perfectly optimized for nothing.

Consequently, design must operate at the level of the entire system (data flows, feedback loops, behavioral governance, and organizational impact).

Naturally, this redefines the designer's role. Designers no longer just draft screens; they establish the terms of engagement between distinct intelligences. They define behavioral rules, map out mental models for both humans and machines, and orchestrate every touchpoint across all coexisting agents.

8. The ultimate goal: Quality of relationship

The real risk isn't that machines are becoming smart. The risk is allowing this intelligence to take shape without intentional, strategic design.

A poorly designed AI always yields the same results: it breeds confusion through opaque decisions, amplifies hidden biases, and degrades user trust to the point where people no longer know when to believe the system or how to correct it.

On the other hand, an elegantly designed system makes its intentions clear: what it is optimizing for, and what boundaries it will not cross. It adapts beautifully without manipulating, leaving users in control, explaining its reasoning, and asking for input at the perfect moment. It genuinely elevates how we live, work, and collaborate.

This is far beyond basic UX. This is about designing a high-quality relationship between human and artificial intelligences.

Conclusion

Design is far from disappearing; it is elevating. We are transitioning from designing objects to designing systems, and now, to designing relationships between intelligences.

In this exciting new chapter, the designer becomes the ultimate champion of these relationships. They are the ones asking the crucial questions that AI models do not naturally ask themselves: What happens dynamically when this system learns, makes mistakes, corrects itself, and continues to live alongside us?

The question is no longer: "How do we design a great product?" But rather: "How do we design systems that learn while elevating meaning, trust, and relational harmony?" This represents a magnificent economic and social opportunity for forward-thinking brands.

References

• Jean Baudrillard, The System of Objects

• Bruno Latour, Reassembling the Social

• Lucy Suchman, Plans and Situated Actions

• Herbert A. Simon, The Sciences of the Artificial

• Ben Shneiderman, Human-Centered AI