5 ways to unlock a culture of experimentation with gen AI

Innovation usually isn’t a lightning strike. More often than not, it’s the result of careful iteration. The first mobile phone weighed more than two pounds with just 30 minutes of talk time. Years of experimentation and iteration produced today’s lightweight, pocket-sized computer/camera/phone combos — and they’re continually evolving.

Generative AI can help organizations of all sizes build a culture of experimentation that drives innovation. Tools like Gemini allow employees to quickly try new approaches, learn from their mistakes, and keep pushing the boundaries of what’s possible. By helping people build on each other’s discoveries, gen AI can be a major force multiplier.

Here are 5 ways to unlock a culture of experimentation with gen AI.

1. Create the right mindset

Turn “failed” experiments into learning opportunities. Consider setting up an internal forum or Google Chat space where your teams can drop in “dead end” prompts or AI misfires, with lessons learned, to reinforce the idea that iteration ultimately drives success. Naturally, you also want to share the big wins when a team uncovers a genuine breakthrough. Some teams at Google send out a weekly digest of AI wins and lessons learned for the week. (Learn more about creating a Google guides program to boost AI adoption.)

2. Invest in education, training, and cross-team collaboration

A McKinsey study found that 48% of employees rank training as the most important factor for gen AI adoption. Yet less than half feel they’re receiving moderate or less support. When Avery Dennison — the world's largest base of Gemini pilot testers— brought AI to thousands of employees, they focused on enterprise-wide training and “humans in the loop.” Their Digital Workplace Services team created a training program and executed a change management plan to upskill both IT and non-technical employees. They used a mix of development days, user training, and a regular stream of AI tips to help ensure all employees were empowered to take advantage of the new technology. From there, the company held functional interviews to brainstorm the best business use cases that could benefit from AI usage.

3. Identify the right workflows and use cases before scaling

Building the muscle for experimentation takes time, even when AI is embedded within your workflows. Some AI usage involves a small, incremental change. For example, writing better and more relevant emails with nudges from Gemini is something all employees can do to boost personal productivity. But when it comes to larger-scale processes, like contract approvals or customer service communications, it’s important to identify the specific use cases and places in the workflow where AI can make a difference.

Based on submissions from their employees, Avery Dennison’s Digital Innovation Center of Excellence created 21 gen AI pilot projects designed to improve productivity for all users and enhance business processes, including maintenance, operations, customer engagement, and workplace safety. For example, by applying AI to predictive maintenance in one of the company’s India plants, the company was able to reduce unplanned downtime by 25%, while improving operational schedules and customer satisfaction and decreasing maintenance costs.

4. Create recipes for experimentation

Some teams struggle to work out when to turn to AI when tackling a specific problem. Consider putting together recipes that guide them through each step and the relevant tools within a hypothetical or real business use case. Unlike blueprints, recipes allow for substitutions and tweaks that make the results more targeted for a specific team. The recipe might give sample prompts and typical responses, while guiding teams on how to get the best out of AI while navigating research, brainstorming, analysis, generating insights and more.

5. Evaluate results, create feedback loops, reinforce safety in experimentation

Periodically check in with teams on the progress they’re making with gen AI. You might send out an anonymous survey or schedule a feedback session that should always include both wins and lessons learned. Evolving your use of gen AI relies on tight feedback loops — not only between the individual user and Gemini, but between your broader AI community and across the whole organization. And as you celebrate building a culture of experimentation, reinforce the ground rules for using AI tools that keep your organization’s data private, secure, and under your control.

Empower gen AI innovation

Creating a culture of experimentation with gen AI can unlock innovation, but it’s not as simple as “turning on the AI switch.” Working closely with users and stakeholders to identify the right use cases, investing in training, and sharing information and resources are all part of laying the foundation for the “humans-in-the-loop” approach that brings out the best in AI.