Ad Agencies And AI: Finding The Balance Of Efficiency And Ethics

Article by Jill Collins, PhD

This article originally appeared on Forbes

Using artificial intelligence at advertising agencies isn’t just an exciting future — it is already happening. But considering how AI has impacted other industries, will this change be for the better or worse? For example, undisclosed AI-generated articles have made their way into major news publications, threatening the integrity of journalism and serving as a warning call to consumers and journalists alike to be more vigilant in the creation and consumption of content.

Additionally, the promise of AI’s benefits is seemingly contradictory to how current agency business models run. Should agencies and/​or their clients be concerned about the increasing use of AI in the industry?

As the CEO of an agency specializing in health, science and wellness, I believe addressing concerns around AI starts with a transparent conversation. So, I want to share how we leverage AI at our agency, and how others can find their own balance between efficiency and ethics.

Overarching Policies on AI Use

Across the board, our agency is guided by policies on best practices and acceptable use of generative AI tools. These include but are not limited to:

• Validating any information or output from a generative AI tool to ensure accuracy

• Never inputting confidential, proprietary, personal or sensitive information

• Keeping an open conversation about AI use and continuing education

I think any agency that wants to successfully and safely leverage AI can benefit from following policies like these.

Here’s a more detailed breakdown of how our teams use AI, and how others can find their own balance.

Designers: Incorporate AI Intentionally in Workflows

Our designers use generative AI the most frequently because it can drastically reduce time to results. For example, they often rely on stock imagery or imagery provided by clients. If an image needs to be resized for visual balance in a composition, generative AI can extend the background seamlessly. Good prompts are key, and outcomes may still be hit or miss. In addition, our designers validate all images for quality and accuracy.

Our design team does not use AI to generate people, living things, objects (e.g., a microscope) or anything that would compromise the authenticity of realistic scenarios. They leverage it strictly for non-consequential elements that do not impact real representation or the storytelling of the image.

Copywriters, Strategy And Research: Explore Use in Limited Capacity

Our copywriters have found that ChatGPT is not yet able to produce high-quality, scientifically accurate content. Instead, our team uses it as a lexicon to generate a quick list of words that meet prompt criteria (e.g., initial sound, syllable length, word meaning). This type of search can act as a catalyst for an ad concept or to find a word that fits perfectly in a headline. Our strategists have also leveraged ChatGPT similarly when developing brand stories and attributes.

For the strategy and research teams, AI has also been useful as a quick aggregator of information for secondary research. Given client privacy concerns and the potential for generative AI tools to hallucinate, using AI is only meant to be a starting point for an initial understanding of the competitive landscape or industry. In addition, AI is not as valuable for clients who are innovating in a new or yet-to-be-established space.

HR, Media and Accounts: Keep it Real

For both our human resources and media teams, AI is often presented as an upsell or offering for optimization. In HR, there are many AI-powered software offerings to help streamline hiring and recruitment. However, due to concerns over biases in the technology, we continue to opt for manual recruitment practices to ensure that every applicant is reviewed and not unfairly screened out.

On the media side, most ad platforms offer some sort of in-platform AI tool for optimizing campaigns. These tools may make suggestions based on budgets or other factors for campaigns. While the team keeps up with the changes and trends inherent to platforms (e.g., what was once manual ​“text ads” is now Google responsive search ads), they still prefer to have more control over campaigns rather than being pushed toward adopting a fully autonomous, AI-driven campaign.

On the accounts side, if a client gets an email from one of our account managers, there is a 100% chance that it was written by a human. That’s because our accounts team does not use any generative AI and likes to keep interactions personal. However, they do leverage various automation tools for task management to help keep their workflows running efficiently.

The Future: Driven By Humans, Aided By AI

AI has made major strides in technological capabilities in recent years. It has the power to change the medical landscape, but at the same time, products and brands that highlight being AI-powered are turning off consumers.

If you are looking to harness the power of AI for your team while also being mindful of ethical adoption practices, here are a few key considerations:

1. Start a conversation. Whether it’s with a team or client, talking about AI can help demystify the process and ensure everyone is aligned on goals and expectations.

2. Keep up to date with changes. MIT Technology Review reports regularly on the topic of AI. Additionally, you can look for team-specific continuing education opportunities from professional organizations.

3. Evaluate your team’s needs first. AI may promise certain advantages, but that doesn’t mean your team will find those same benefits.

4. Set parameters with policy. Establishing a framework through policy can protect your team and your clients. It can be updated over time as AI practices evolve.

In my opinion, being an ethical adopter is far more important than being an early adopter. Our agency spans a full spectrum when it comes to adopting generative AI — all with an eye toward making responsible, ethical choices in how to leverage these tools. AI may have its place in helping synthesize data and drive efficiency, but it doesn’t share in the emotions and lived experiences of humans. That’s where the real magic happens.