Embracing Generative AI: A Responsible Approach
Written by Mike Creighton, Executive Director of Experience Innovation
Generative AI has arrived. We don’t have to squint to see its potential for disruption across industries. The disruption is happening in real time. With each day, companies and individuals are discovering new applications, uncovering new efficiencies, and reexamining old problems. We’re doing the same. As an integrated digital agency rooted in technology, our entire company is buzzing with excitement and ideas. We see opportunities everywhere.
We’re no strangers to AI and know it powers many of the digital products we use today. From business transforming vision work to playful experiences, we’ve helped clients think about and realize the possibilities AI offers. We’ve partnered with Google to demonstrate many of their machine learning technologies—whether it’s our TensorFlow powered space-themed arcade or our emotion-detecting photobooth using the Google Cloud Vision API. With others, we’ve thought deeply about how AI could bring more personalized shopping, fitness coaching, and wellness experiences to consumers. We’ve envisioned interfaces and services that can respond to the variability of experiences our current paradigms aren’t capable of delivering.
And, still, generative AI feels different. What was once reserved for vision concepts is now in everyone’s hands. In recent months, we’ve seen a dramatic change in the accessibility of AI along with a realization that these foundational models have incredible capabilities and applications. Across our company, we’re already experimenting with a variety of tools and services, prototyping ideas that will change the ways we operate our business and the ways we as individuals perform our daily tasks. The use cases seem limitless.
But we’re doing all of this with a commitment to thoughtful and responsible adoption. We won’t move forward blindly, so we’ve outlined some steps to integrate generative AI responsibly.
OPENAI GPT-4
GOOGLE BARD
RUNWAY
MIDJOURNEY
AUTO-GPT
MICROSOFT BING IMAGE CREATOR
MICROSOFT BING CHAT
ADOBE FIREFLY
Across our company, we’re already experimenting with a variety of tools and services, prototyping ideas that will change the ways we operate our business and the ways we as individuals perform our daily tasks.
Know Your AI
Understanding generative AI is crucial as it will change many aspects of our business and how we as individuals work. This new wave of generative AI is different from other types of AI we see in our daily lives. It’s a tool that we can manipulate, control, and direct without any prior knowledge. This is a novel shift in our relationship to AI.
To ensure every employee is prepared for this change, we've swiftly rolled out comprehensive training that covers various types of generative AI, their applications, potential risks and ethical issues, and how they relate to our clients and our industry.
This education goes beyond surface-level knowledge. We want our employees to truly understand how these generative AI models work — their potential and limitations. This technology will be transformative and impact all of us, so we want our employees to respect its capabilities and the potential for misuse.
An hour long primer on text-driven generative AI created by Mike Creighton for Instrument employees.
Mindsets Matter
We encourage two key perspectives when using generative AI:
First, AI-generated content is never final; we should always scrutinize and refine it. With LLMs, this approach involves proofreading and fact-checking the output for bias or potential harm. We never blindly accept the generated content as accurate or true. When we’re using text-to-image models, we’re the director, the curator. The output is the inspiration and jumping off point for further iteration and exploration. In both cases, our honed craft plays a key role in bringing the output to a finished state.
The second perspective is that we embrace AI as a collaborator, teacher, and assistant, not a threat to our roles or our value. Generative AI can replace some tasks previously considered exclusive to humans, but it can also help us focus on higher-value challenges and contributions. By seeing AI as our partner, we can unlock new creative potential and enhance our critical evaluation skills.
Inclusive and Equitable by Design
We're committed to ensuring all employees have access to vetted AI tools, resources, and training. We know that generative AI will change the way all of us work. There are new skills to learn and wholly new ways to approach our problems. So we’ve created an always-on centralized repository of resources to ensure that our employees don’t stagnate and have the tools for building these new skills. We’re coupling this with ongoing programming including panel discussions and demos from representatives of all our disciplines to illustrate how everyone can use these technologies.
Diverse perspectives amplify our success with AI integration. We’ve begun an inclusive process for vetting new tools and services that involves employees from different backgrounds, disciplines, and seniority to maximize opportunities and uncover scenarios where bias, harm, or ethical issues may come into play. We recognize that this has to be a group effort so that we don’t overlook anything.
Serving Clients Better
We work with some of the world’s largest technology and service companies — companies that are creating products for humans around the world. In many cases, AI is driving transformative features and utility in their offerings. As good partners, we see it as our responsibility to deeply understand these technologies and bring a necessary perspective that ensures inclusive, accessible, bias-free products.
We're also committed to data privacy and confidentiality. With a number of publicly available generative AI services, input data isn’t private. It can be reviewed by their employees, used to further improve and train models, or even be made transparently available to the public. So we’re educating our employees about these cases and have rolled out strict policies around acceptable use for input data. We'll only use confidential information as input for generative AI when a service meets security requirements and complies with our clients' and our own data privacy policies.
We’ve begun an inclusive process for vetting new tools and services that involves employees from different backgrounds, disciplines, and seniority to maximize opportunities and uncover scenarios where bias, harm, or ethical issues may come into play.
Moving Forward
As we continue experimenting with generative AI in daily practices, we'll uncover risks and issues that warrant new principles, policies, and guidelines to ensure ethical and responsible use. Open, critical dialogue among our employees, clients, and the public will drive this process. While we’ve already defined some foundational elements around policy and best practices, there’s still more to do, and it’s some of the most important work ahead of us.
We’re eager to continue exploring the possibilities this technology has to offer, and by taking the time to understand it and create the appropriate safeguards, we’re ensuring generative AI benefits our employees, our clients, and their consumers. We create work that touches the lives of so many humans. As a company whose purpose is to shape a better future, we are committed to moving forward responsibly and thoughtfully into this new era of AI.
