Nicki Sprinz is the CEO of ustwo, an award-winning, majority employee-owned experience design agency.

Optimizing for efficiency worked in the relative stability of the 20th century. But now the world’s changing quickly, so if you solely pursue efficiency, you risk becoming brittle because it will be harder to respond to what’s coming next.

Being effective in today’s world means letting go of old certainties and stepping out of your “business as usual.” Adaptability is critical. Organizations need to be responsive to a constantly changing environment—technology changes, but also demographic, political and climate changes.

In the context of accelerated change, it’s essential to explore new technology that might make a difference, like artificial intelligence. But you have to do it in a way that is responsible and builds trust. Here’s how to do that:

Explore AI, But Keep It Aligned With Your Values

Experiment with AI, get curious, measure results, and explore the effects.

And consider the questions that using AI brings up: Does it create new opportunities? Can the output be trusted? Was the data responsibly sourced? Is the AI extractive or exploitative? Who benefits from and who is damaged by your use of AI? Does AI accelerate your work, or does it completely change that work?

Whatever you do with AI, your company values should guide your decisions. When my company works with AI, we need to understand the why behind AI—the intentions of our clients. We don’t work on products that are designed to lead to redundancies, and we’re not trying to build AI unless there’s a real need. It’s a tool, not a product. We keep people at the center by making breakthrough digital experiences that are about people and outcomes, not the technology. These values are a foundation; they inform and create guardrails for our use of AI.

Effective AI: Responsible, Transparent, Sustainable And Flexible

We think responsibility and trust underpin using AI well. On the ground, that means ensuring that your use of AI is responsible, transparent, sustainable and flexible.

Responsible:

• Ensure your data is ethically sourced and inclusive. People should be aware that their data is being incorporated into a training dataset, and the data should come from a diverse group of people. We’ve already seen negative unintended consequences of algorithmic bias. If you’re not building your own data model, make sure the third-party model you’re using adheres to these principles.

• Review and validate your training data and reference sources throughout the process.

• Provide an impact analysis to internal teams and communities, taking into account the long-term considerations of building an AI model or using AI in a digital product.

Transparent:

• Make it clear when content is AI-generated to maintain user trust. For example, label AI-driven experiences, or design digital experiences with moments of friction that inform the user.

• Explain algorithmic decisions to help assess quality and prevent bias.

• Stay compliant with regulations and keep pace with changes.

Sustainable:

• Reduce energy consumption for cloud, on-device and on-premises workflows with energy-efficient hardware use. For example, move AI processes to systems powered by renewable energy.

• Select AI models that factor in performance and cost.

• Limit expensive AI operations like retraining.

Flexible:

• Build brand relationships through emotive AI (when AI experiences are designed to resonate emotionally with the user). Be sensitive to users’ emotional states and context.

• Avoid vendor lock-in by being agnostic about models and AI infrastructure.

• Work at a faster pace to bring ideas to market: Think of it as experimentation as a service.

Avoiding The Sea Of Sameness

These four principles result in AI that’s more trustable and more responsible because teams and people using AI can see that it’s more trustable.

But to achieve AI that is more successful, you must also create emotionally resonant experiences. A side effect of the rapid growth of big technology platforms is that digital experiences have become formulaic. It’s like a sea of sameness. They are very proficient at serving audiences at a massive scale, but they often fail to resonate: Users don’t stick with them.

So AI-powered services must offer an emotive, engaging experience and stimulate an emotional response. Personalization is table stakes now. It’s not enough for AI to know your basic preferences or carry out a task on a functional level—it needs to understand and account for your emotional state. Combining emotionally aware design with the speed and power of AI lets you create services that deeply connect to people.

For example, a successful chat experience might better connect to a customer by being able to interpret when they’re frustrated and adapting the AI’s voice and tone to that emotional situation. An overly chipper bot is probably going to make someone who’s irritated that their flight was changed or their booking lost even more annoyed. Or a financial services digital experience can change how it speaks to a user based on their knowledge and comfort with a topic.

As AI experiences become a bigger part of our lives, like any relationship, they require trust—and only by designing in a responsible way can we maintain user confidence and loyalty. And like people in relationships, AI agents need to understand us. Experiences that lean on emotive AI, taking the tech beyond functionality to resonate with users on a deeper level, are going to come out on top. Sticking to our values and humanity in an AI-driven world is ultimately a good business practice, too.

Forbes Agency Council is an invitation-only community for executives in successful public relations, media strategy, creative and advertising agencies. Do I qualify?