Generative AI Manifesto

Impulse Creative's Generative AI Acceptable Use Policy 

NOTE: If you're an agency that is creating a similar guide, we'd love to collaborate. This is a new frontier and we would love to continue to build this manifesto to better serve our clients and the ecosystem as a whole.
Please contact us to talk more about Gen AI

Purpose

The ultimate goal of any agency is to help its clients grow, but that relationship must be met with four core focuses: anticipating growth opportunities, creating (more) value, creating less work for clients, and maintaining confidence. 

Generative AI has exploded and become more accessible than ever in 2023. As the Generative AI landscape advances, agencies face a crucial decision regarding using Generative AI to service clients, but it’s not their decision alone: clients must also weigh in on how Generative AI is used in pursuit of their own growth. While agencies are hired for their ability to shorten time-to-market and reduce the risk of hiring new teams, Generative AI represents a potential game-changer in the industry but comes with potential hurdles that must be addressed ahead of time and overcome. 

There are still concerns about whether clients will see the same level of value and confidence in their agency partners when AI is used. While Generative AI is still in the “innovation stage,” agencies should be “early adopters” who properly vet solutions to guide companies' growth journey.

As the industry evolves and new solutions emerge, agencies must navigate this crux with care to ensure they are meeting their client’s needs.

This manifesto defines how we strive to leverage Generative AI and emerging technologies in general. But also to ensure clients have the right expectation and understanding of how agencies believe this innovation will affect their business. 

Ethical Servicing

Trust acts as the wind in the sails of any agency-client relationship, and just like sailors tack into the wind, agencies have to navigate with care: clients expect their agency partners to help them reach their goals by completing work as scoped and specified within timelines and budgets.  The proliferation of AI-driven technology, both standalone solutions such as ChatGPT and solutions built into tools such as HubSpot and ClickUp, are compelling options that could help agencies meet their client’s expectations. However, agencies and clients may use these nascent tools without fully understanding the potential effects on performance, accuracy, copyright, intellectual property, and many other factors (including trust).  This means that when using Generative AI on a client’s behalf, agencies must understand both the ethical implications and the service quality risks of replacing humans with machines.

To address this, agencies must proactively seek their clients' agreement and inform them of their intentions of using AI-powered tools. The focus should be on testing, vetting, and validating the use of Generative AI to identify its benefits and drawbacks. Agencies must also ensure clients understand how incorporating AI into their businesses impacts service, quality, delivery timelines, and overall costs–and yes, when restructuring projects to use Generative AI, agencies may need to restructure their fees.

Clients want to ensure they get the most value from the human aspect of service delivery with or without using Generative AI. Therefore, agencies must adjust their timeline and costs to include AI's impact on the time taken to execute creative work accurately if they choose to use Generative AI. Failing to balance these aspects will lead to ethical problems in the agency-employee and client relationships. 

... leveraging Generative AI could potentially cut content creation times by up to 30%, resulting in an increase in output by the same margin. However, platform costs and potential quality ramifications must be considered. ChatGPT by OpenAI has a negligible or zero cost at present, but the reality is that you are the product, and there are potential ramifications of ownership of the IP.

The service model might change if Agency and Client mutually agree to leverage Generative AI. However, most Generative AI companies have not yet fully defined their cost models, posing a challenge for the early majority. Future dependencies on the platform and subsequent changes in overall costs could also emerge. As such, agencies face the critical task of determining the costs of embracing the new technology and comparing it with how it will improve production. 

Explicit Client Notice & Explanation

As Generative AI becomes increasingly widespread across various departments, clients must be informed explicitly of its use on their behalf. Generative AI involves leveraging algorithms and machine learning to create editorial content, graphic design, and more. Agencies should be both ethically and contractually bound to inform clients that their data may be used to train these algorithms, thus resulting in AI-generated content that may not always be perfect. That's why human intervention is still a critical need to ultimately mitigate potential biases, misinformation and validate.
Therefore, agencies must be transparent with their clients about their use of Generative AI and provide clear explanations of its purpose and limitations.

This not only develops trust with clients but also promotes the responsible and ethical deployment of emerging technologies.

Data Security of Client Information

Agencies act as customers' agents, and as ethical entities, they typically include a confidentiality clause in their contracts. While these clauses may not explicitly pertain to AI, it is incumbent upon agencies to teach their team members about the potential implications of incorporating client data into learning models for Generative AI. The stark reality is that confidentiality may be breached, and many companies remain oblivious to the dangers associated with installing "plug-ins" or utilizing ChatGPT or other Generative AI models that contribute to the learning model.

Companies such as HubSpot are including disclaimers in their products to notify of potential confidentiality concerns in using AI..

Example from HubSpot: Please avoid sharing any sensitive information in your prompts. To improve the product, HubSpot logs and stores your prompts, generated language, and usage metrics when you use AI content assistant. For content moderation purposes, HubSpot shares your prompts with OpenAI, and your prompts will be attributable to you.

Granting explicit permission for each AI tool/model is necessary for employees, with vetting by leadership and sign-off by clients. Not taking this explicit step may put agencies in breach of their contracted confidentiality agreements and data security/privacy agreements. Agencies should also document and share the outcomes of these conversations.

When approached with the option to use these tools, clients should consult with their data security and/or compliance teams to understand the usage profile on their accounts, and ensure that the solutions align with their needs.

Legal & Copyright Considerations

As Generative AI becomes more widespread, agency-client relationships must be aware of various legal and copyright considerations to avoid potential legal conflicts. Copyright ownership is a key issue, as Generative AI often uses pre-existing content to generate new works. 

This raises questions over whether the agency, the client, or an original creator owns the copyright of the final work and whether they have obtained the proper permissions to use the pre-existing content. 

Moreover, it is important to ensure that the new work does not violate any intellectual property rights or trademarks of other parties. An example of this would be the recent copyright claim of Zayra of the Dawn where the US Copyright office granted copyright to the book as a whole but not to the individual images in the book, claiming that the artist did not sufficiently produce these images as Generative AI created them.

Another consideration is the impact on data privacy regulations, as Generative AI relies on large amounts of data to generate outputs, which may require consent from individuals or businesses. 

Only after understanding and addressing these legal and copyright ramifications, should agencies and clients leverage the benefits of Generative AI in their own marketing campaigns, client deliverables and other public-facing assets. 

It is also worth noting that the evolution of acceptable usage and ownership rights of AI-generated content is ever-changing, with countries such as the U.S. and within Europe exploring more formal regulation. This could have significant implications, particularly for agencies who typically transfer content ownership to clients upon payment. Stay alert, as changes in this space are happening fast.

Platform Adoption

Increasingly, many tools that agencies and companies are already using (or considering using) are adding AI to enhance the performance and productivity of their users. We break down these tools into three main groups:

1. Point Solutions

Tools that enhance their feature set with Generative AI capabilities as an add-on (whether included in the subscription or additional)

  • HubSpot
  • Canva
  • Grammarly
  • Photoshop
2. AI Platforms

Tools which their primary focus are for generative AI purposes. Specifically around Content or creative works. 

  • Jasper.ai
  • Midjourney
  • OpenAI’s Chat GPT
  • GitHub Copilot
  • Fraze
  • Adobe Firefly
3. AI Data Processing
  • Analytics tools
  • Opteo (adwords)

Before adopting these tools or add-ons for clients, agencies should begin by experimenting with these features for their own needs. This ensures solutions are properly vetted and understood in a safe and ethical way. Any employee involved in this vetting period must consider acceptable use, ethical servicing, and potential client results. At no point should an agency (or their employees) use these tools for clients without disclosure and keeping in mind the other covenants of this document.

Cost Considerations

When adding Generative AI Platforms that can be added to the tech stack, organizations should take special care and attention to understand the expected costs of the solutions. If the solution is free, it’s important to remember that it’s likely that “YOU” (namely, the data or information you provide the tool) are the product. In addition, as Generative AI platforms and add-on features continue to evolve, it’s to be expected that pricing will as well.

Where possible, Generative AI platforms should be launched “on behalf” of clients where the client owns the account, and the agency assists in leveraging it.

Applications of Generative AI

Individual use cases will dictate the endless applications of Generative AI, but as a beginning of the exploration on how it can tie into agency operations, the following concepts should be considered: 

  • Generative AI should be used to enhance content creation, not replace it.
  • Agencies can tap into the power of Generative AI for topic and concept development. This enables them to use strategic prompts and obtain a diverse set of results. These results can then be used to obtain client feedback and direction and boost the efficiency of the creation process.
    • Agencies should vet any content created and ensure proper (and true) attribution; fact-checking is essential.
  • Generative AI can be used to enhance imagery for use in different formats (see Adobe Firefly) such as to create specific and relevant graphics for use with Social Media or supporting material for websites.
  • Generative AI can assess content to maintain a certain tone of voice and brand standards.
  • Generative AI can be used to vet and assist in data management (keep mindful of privacy and confidentiality). It can be used to gain insights into data depending on the platform.
  • Developers and software engineers can leverage tools to assist with debugging existing code in the development process
  • Agencies can use Generative AI to analyze user data and preferences to offer personalized recommendations for products, services, or content. This can enhance the user experience, increase customer loyalty, and drive sales.
  • Generative AI can aid agencies in forecasting future trends, patterns, or events based on historical data. This can help agencies anticipate future challenges, allocate resources, and make better decisions. 
  • AI driven-chatbots can be built to provide instant customer support, answer FAQs and provide personalized solutions. This can increase customer satisfaction, reduce costs, and speed up response times.
  • AI can power moderation for community forums or conduct sentiment analysis to help organizations understand their users.
  • Natural language processing (NLP) technology can be used to generate automated reports from large bodies of text. This is useful in uncovering trends or patterns that would otherwise go unnoticed (e.g. Chatspot Alpha)
  • Generative AI can be used to repurpose content from one format to another for different mediums, such as recording a webinar and using AI to turn the transcript into snippets for sales teams or blog topics for content marketers to write.
  • Generative AI can be used for script writing for podcasts or videos by helping organizations reformat their thoughts or existing work into content intended to be spoken.

Generative AI & Strategy

There's a big emerging misconception about how Generative AI can be used for "strategy." A quality agency should provide not only outputs but also strategic direction. Generative AI can support this direction, but blindly accepting its "strategic campaign building" is a mistake. While it may provide a starting-point it’s important to not rely solely on its output for your strategy. 

For creative campaigns, context is key. The best output is often derived from understanding the target audience's behavior and interests, as well as leveraging historical data or insights. AI can help sift through this data to find interesting angles for your go forward, but is not a replacement for a strong strategy based on analytics and a clear understanding of objectives.

Finally, don't forget that human creativity is still required. AI can provide numerous options for creative campaigns, but a human touch is essential to refine the content to be effective and unique.

It’s important to recognize that AI has much to offer to improve efficiences and diversify ideation, but human intervention is necessary to achieve the best results. Strategic direction should be provided from the outset with a strong focus on understanding consumer needs, organizational goals, historical insights and more. 

FAQs

Employees are representing and working on behalf of an agency on client work. As per contractual obligations, this typically includes confidentiality and ethical responsibilities.

The use of Generative AI without disclosure to the agency (and such the client) to execute work may be grounds for termination. Just as an agency has to navigate the ethical implications of leveraging Generative AI without explicit notice, employees should be held to a similar standard.

 

As of July 2023 we do not offer formal service offerings that include Generative AI. We are currently vetting partnerships with Generative AI platforms such as Jasper and are keeping a close eye on platforms like HubSpot & ClickUp that are adding in Generative AI capabilities. We are actively testing platforms and tactics on ourselves.

If you are an existing client, feel free to ask our team where and how we best think Generative AI can be used.

If you're interested in becoming a client, please reach out to us and we will be happy to talk about options.