Image background generated in Midjourney

Craft’s Principles for Good AI

At Craft, we have a simple philosophy; Good Work with Good People for Good Clients with Good Balance. To ensure we’re doing that good work, we put the people who use a product at the center of the product design process, designing solutions rooted in our deep understanding of their goals and the obstacles that stand in their way. As we help our clients unlock the full potential of AI, we think it’s important that we extend our philosophy to ensure we are designing products with Good AI.

To develop our point of view on Good AI, we looked to work we have done with our clients for inspiration and surveyed our team of designers, researchers, and engagement professionals to better understand how we use AI, what we hope for it, and what concerns us. Using this information, we ran a series of education sessions and workshops with our team to begin framing our perspective.

These Principles for Good AI are an initial expression of our evolving understanding and cautious optimism about the potential of AI. We recognize its potentially immense benefit to people through its additive power across a multitude of applications. We balance that with recognizing and understanding the risks of overfitting training data and promulgating biases at a large scale. These principles underpin the responsibility we take as User Experience professionals to ensure these capabilities are built to support, not disenfranchise, the people impacted by them.

Principles for Good AI

1. Good AI solves real problems.
   AI should meaningfully and measurably address a clearly defined problem. It should not be a novelty capability without a valid use case.
2. Good AI unlocks human potential.
   AI should elevate the abilities and the work of the people who use it, enabling them to achieve far more and far greater than they could have on their own.
3. Good AI is easy to understand and control.
   AI should interact within the context of the people who use it, adapting to how users engage and providing guidance to help them best leverage the capability.
4. Good AI is for everyone.
   AI should be inclusive, adaptive, and accessible to all, irrespective of their background, beliefs, ability, identity, and orientation.
5. Good AI is additive.
   AI should exist to supplement people’s efforts, not replace them.
6. Good AI is transparent.
   AI should provide clarity on where it pulls information from to generate its responses to enable people to determine how to responsibly use the results.
7. Good AI builds trust.
   AI should provide transparency into how it is trained and be flexible to evolve using data sets that are representative of the real world to minimize the risk of bias.
8. Good AI is collaborative.
   AI should help people to interpret the outputs they receive and determine how they should best proceed.

Good AI is our responsibility

A common thread across our AI principles is the need for people to interact and collaborate with AI with intention. When decisions have the potential to impact us, we should not be handing over the reins without question. Our role in the creative and decision-making process with AI creates a system of checks and balances. We should be interrogating the models that drive AI, evaluating the outputs with an inquisitive eye, and accessing any information that can help us form an opinion on our confidence in these outputs. In the end, AI is a tool that must be wielded in the hands of informed users to ensure that it delivers a tangible benefit to people rather than disenfranchise them.

Learn and collaborate with us

These principles reflect our initial thinking on AI’s potential. This thinking is based on how our agency has used these capabilities and designed for them so far, coupled with our deep expertise in human-centered design. We fully expect that as we learn more, these principles will evolve. As we intend to grow our knowledge, we value the input of diverse perspectives. Our hope is that these principles spark dialog with our clients, industry leaders, AI enthusiasts, and AI skeptics in a way that helps us further refine our point of view. If you have feedback, a reaction, an idea, or even a counterpoint based on what you read here, please let us know at hello@madebycraft.co. We’d love to hear from you.

Special thanks to Sean Neville, Matt Venables, Michael Burke, and Sam Zimmerman for their insights and perspective and to the entire Craft team for their curiosity and thoughtfulness.