Skip to main content

AI Policy

“Smarter” isn’t the end game. Better is.

 

At Clarity, we support a lot of high-tech innovators. Our clients are most often organizations who make sophisticated devices, apply cutting edge techniques to familiar problems, engineer more capable software, or all of the above. What always draws us in is the story of how our clients’ innovations will have a real and immediate impact on improving human lives.

Our own field is now experiencing rapid innovation. Leaps in generative AI have made software smarter, fast. With that change comes incredible opportunity and possibility. But we’re also at a crossroads. Critically considering how and when to use generative AI – and when not to – will determine where this transformative technological shift takes us, as individuals, as organizations, and as a society.

The ideal is that AI enables all of us to work more intelligently and effectively. By doing so, we can shift time and energy to tasks that are fundamentally human in nature – building connection and community, and guiding further innovation based on our deep understanding of those communities. Reaching that ideal depends on responsibly applying AI.

 

This statement clarifies our principles for responsible and ethical AI use at Clarity CPH.

We invite you to use this document, in whole or part, under a Creative Commons Attribution-ShareAlike license as a starting point for your own responsible AI policies and practices. This Creative Commons license lets others remix, adapt, and build upon the work, even for commercial purposes, as long as they credit the source and license their new creations under the identical terms. That’s how we got started, springboarding off the work of the smart folks at Marketing AI Institute.

 

 

Our Responsible AI Principles

  1. We believe in the responsible design, development, deployment, and operation of AI technologies.
  2. We believe in a human-centered approach to AI. We believe that AI technologies should assist and empower humans, not replace them.
  3. We believe that humans must oversee and remain accountable for AI-assisted decisions and actions.
  4. We believe human knowledge, experience, emotion, and imagination is essential for effective communication. We believe that maintaining and using that expertise is critical as AI becomes part of our workflows.
  5. We believe in the power of language, images, and videos to impact individuals, groups, and societies. As such, we commit to never knowingly use generative AI technology to deceive; to produce content for the sole benefit of financial gain; or to spread falsehoods, misinformation, disinformation, or propaganda.
  6. We believe in continually educating ourselves on the changing limitations, dangers, and legal implications of generative AI, and considering those factors in our recommendations, decisions, and actions.
  7. We believe that transparency in AI usage is essential in order to maintain the trust of our audiences and stakeholders.
  8. We believe in personalization without invasion of privacy, including strict adherence to data privacy laws and protection of confidential client information. When legal precedent lags behind AI innovation, we are committed to following our moral compass.
  9. We believe widespread AI literacy is essential to mitigate the potential risks of AI. Therefore, we are educating ourselves, and we are taking an open approach to sharing our AI knowledge, ideas, experiences, and processes to support others navigating these complexities.
  10. We believe in partnering with organizations and people who share our principles.

How We Use AI Today

Because generative AI is changing and growing so rapidly, the information on this page is not static. As we evolve, we aim to keep you updated. Consider this page a snapshot of where we are.

We approach AI tools the way we approach business in general: we always want to give our clients the greatest value we can for their investment. That means we select both our specific AI tools and the ways we use them based on our best understanding of

  • what our clients need,
  • what we can see the tools do well, and
  • potential risks (e.g., copyright infringement or limitations, exposure of confidential client information, etc.).

As a result, we currently use a collection of standalone Software-as-a-Service (SaaS) generative AI products (e.g., Perplexity AI, ChatGPT, Adobe Firefly, Playground AI) and integrated generative AI tools (e.g., Microsoft Copilot, Adobe Generative Fill) experimentally or regularly for the purposes below.

Business purposes

  • Speech-to-text transcription of meetings
  • Summarization of meetings
  • Assisting in writing formulas for spreadsheets

Copywriting purposes

  • Initial research and information gathering. Generative AI can function much like search engines do: enabling us to quickly identify interesting directions and key information when we need to learn about a product category, market trend, or the mindset and job responsibilities of a specific professional role in our target market. Just like we do when using a search engine, we always cross-check information we intend to use with vetted and trusted sources.
  • Outline generation. For pieces such as slide decks, white papers, and ebrochures, we may experiment with using AI to generate a list of topics to be covered. We may use this list as a starting point to revise and build on, or as a comparison tool to help us evaluate our existing plan.
  • Sparring and brainstorming. For example, we currently generate lots of headline possibilities manually, which we then evaluate, revise, and refine using our knowledge of the product, the communication problem to be solved, and the audience. Using an AI tool can speed up the idea generation part of this process without substantively changing what we do.
  • Rapidly generating A/B test options. Because of how AI speeds up idea generation, we are interested in experimenting with it to create A/B test options. We may use this internally – for example, asking the AI to use a white paper we have written to suggest different structures for a 2-page flyer. In addition, we may give the AI small samples of text we have written (such as social media posts, headlines, and taglines) and prompt it to provide new suggestions as part of the brainstorming and sparring process to come up with additional options for our client.

Note: We never use generative AI to create anything close to final text for clients. Although we’ve found AI helpful in supporting brainstorming, ideation, and organizational tasks, we believe humans are best at talking to humans. In addition, ongoing copyright litigation about how AI models were trained, as well as uncertainties about when or whether AI-generated works will be eligible for copyright, pose too many potential risks for our clients.

Visuals

  • Brainstorming, sparring, and ideation. We always develop our Creative Concept ourselves, using our own Communications Strategy as a foundation. However, once we have a few potential Creative Concepts for a project, we may use AI internally to rapidly explore the potential of the concepts and develop them further.
  • To create mockups that help clients visualize and understand the concept for a photoshoot, or to demonstrate how we might modify a licensed image from Getty or iStock to fit a client’s needs.
  • Photo editing for mock-ups. For example, extending or editing backgrounds, removing or adding objects, adding effects.

Note: We do not use generative AI to create final images for any of our clients.* Ongoing copyright litigation about how AI models were trained, as well as uncertainties about when or whether AI-generated works will be eligible for copyright, pose too many potential risks for our clients. Finally, when we use AI for mockups, we always disclose that to our clients.

* We cannot guarantee that stock images are not AI-generated. We have already seen AI-generated images in some mainstream stock image libraries. However, we ourselves will not knowingly generate final images using generative AI, and we will always inform the client when we believe a stock image is, or is likely to be, AI-generated.

 

Our playground

Creative agencies are full of creative folks. It’s no surprise that we like to play and experiment.

In addition to the use cases described above, we have a sandbox where we play with generative AI in other areas. Currently, that’s mostly for sparring, brainstorming, and ideation for various other media, such as storyboards, video concepts, and animations.

About This Document

This is v1 of our Responsible AI Statement.

Our hope is that it inspires others to think more critically about how they are using AI technologies, and put responsible AI principles and policies in place to guide their teams moving forward.

We welcome feedback from our community, and we will share future updates in the same open-source spirit of v1.

All parts of the statment can be used under the Creative Commons license below.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.