Agencies weigh in on AI marketing

Hello marketers. Welcome to AI Marketing School, where we dish out the latest and greatest in AI-powered marketing.

In this week’s issue:

1. AI Marketing Update: Agencies give away their AI workflows

2. The Stack: GPT-o1 - what is it good for?

3. Consultant’s Corner: The perils of marketing AI products

Onwards!

AI MARKETING UPDATE

How agencies use AI

Learning from examples is a great way to build and refine AI strategies.

Agencies are busy integrating AI into their daily operations, and there are some fascinating workflows on offer.

Let's take a look at how some of the most innovative minds in marketing are putting AI to work:

1. Billion Dollar Boy: This influencer marketing agency set up Muse, an emerging tech arm that uses AI to visualize fictional characters blended with real-life influencers. They use tools like MidJourney to create concept images that help brand partners understand ideas more easily. "It's all about communicating intangible ideas in our heads to brands on a screen," explains Henry Crisp, a senior creative. (Credit)

2. Gut: The agency is training a "promptless" AI tool to process and combine data sources, creating AI personas designed to model people in different life stages and backgrounds. Brands can solicit input and advice from these personas based on their target demographics. Christian Pierre, Chief Intelligence Officer, calls it "kind of like having a focus group on standby." (Credit)

3. Mischief @ No Fixed Address: This agency is using AI tools like Waldo and Jasper primarily for internal materials, such as imagery for decks, rather than for public-facing content. "We're double-checking everything that we see," says Darrin Patey, VP of Creative Technology, emphasizing their focus on maintaining quality control.

4. Interstate Creative Partners: AI has transformed its mockup process, allowing the team to generate imagery that fits specific aesthetics or briefs in seconds rather than hours. Matteo Di Iorio, Associate Partner, notes: "AI is a tool we value for its ability to bring to life hard-to-define ideas and use them in concepts we can reference for our work."

5. Poppins: The agency uses AI to visualize concepts for clients, particularly in situations where the client has the product but not the scene. They use Midjourney to generate 3D scenes for placing designs, from a hand holding a mobile phone to Brutalist architecture for car placements. Dan Sherratt, VP of Creative and Innovation, explains: "This speeds up our communication with clients to get projects off the ground."

6. Wonder: They're using AI to unify tone of voice across collaborative work. Creative Director David Crease explains: "By simply passing this copy through ChatGPT, including a prompt like: 'make this copy succinct, punchy and with the energy and tone of Tony Stark'’ we can create a unified output." This approach helps maintain consistency in multi-author projects.

And this is just scratching the surface. Google Cloud recently compiled a massive list of 185 real-world AI use cases across numerous industries. All Google-related, of course, but nevertheless an impressive compilation.

One thing that’s clear from these examples is that while there's no one-size-fits-all methodology or best workflow, the common thread is letting AI handle the grunt work and free up human creativity for higher-level thinking.

THE STACK

The art and science of mastering GPT-o1

OpenAI’s new GPT-o1 has already been out for a few weeks, but it’s fair to say that many are scratching their heads about precisely when or how to use it.

If you’re not familiar, GPT-o1 is designed to ‘think’ more deeply before responding.

Here’s a brief summary of what it does:

1. Chain-of-thought reasoning: GPT-o1 breaks down complex problems into logical steps, showing its work along the way.

2. Enhanced problem-solving: In testing, o1 outperformed expert humans on PhD-level science questions and solved 83% of International Mathematics Olympiad problems – a massive improvement over GPT-4o's 13% success rate.

3. Two versions: o1-preview for complex tasks, o1-mini for quicker, simpler queries (but still more advanced than GPT-4o for certain problems).

4. Increased cost: Approximately four times more expensive to use than GPT-4o due to its complex reasoning processes.

I tested GPT-o1 with a very tough task, asking it to “Ponder this study on the Big Bang and posit how it conflicts with existing theories. Use multiple steps to analyze the text and conceive your arguments.”

The context: A recent study challenges the Big Bang as the leading theory of universal expansion.

The thought process below reveals o1’s ability to consider a complex task far more methodically than other LLMs. Being able to see the workings behind each stage is amazing — it makes it easier to understand the AI’s output and sparks new ideas for us to explore, too.

So, how can we use o1 for marketing?

In many situations, it comes down to this: try it when other LLMs (Claude Sonnet, GPT-4o, etc.) are struggling and you’re getting frustrated. 

From personal experiments and scouring Reddit for use cases, this includes situations like:

• Understanding massive data dumps: One great use of AI tools is dumping loads of text and querying it for data, themes, etc. For example, you could dump all of your existing blog content into ChatGPT and ask o1 to work through it more methodically, grouping themes and identifying inconsistencies, gaps, etc. I read someone on Reddit reported much better performance analyzing a 90-minute customer conversation with GPT-o1 than GPT-4o.

• When you need more objectivity: Most LLMs have sycophant tendencies, meaning they sometimes try to ‘echo’ your ideas back to you. For example, if you create a doc with lots of technical concepts and ask GPT-4o to fact-check it, there’s a chance it’ll avoid telling you that you’re wrong. If you need to check something complex, use o1 and compare the results. 

• When brainstorming goes around in circles: Sometimes, AI feels like it’s rushing to simply re-hash a previous response, swapping a few things about it, i.e., it’s just not thinking. o1 is great for thinking outside the box — ideal for brainstorming social marketing strategies, product launches, content ideas, etc.

A couple more things. When using o1, keep in mind:

• It's slower and more expensive than GPT-4o, so use it strategically.

• It can overthink simple queries, so save it for truly complex tasks.

• It lacks some features of GPT-4o, like internet browsing or image analysis.

• You have to encourage it to apply it’s multi-step reasoning

Overall, it’s definitely worth experimenting with o1 if you aren’t already, especially if you’re frustrated with AI’s performance on some tasks.

CONSULTANT’S CORNER

Products that mention “AI” in the description face marketing challenges

An intriguing study in the Journal of Hospitality Marketing & Management uncovered a surprising trend we need to be aware of: simply mentioning "artificial intelligence" in product descriptions can decrease consumer trust and purchase intent.

The researchers conducted six experiments using a mix of fictitious brands and product descriptions. They manipulated AI terminology while keeping other features consistent.

Here’s the lowdown:

1. Negative impact: Including "AI" or "Artificial Intelligence" in product descriptions led to lower purchase intentions across multiple experiments.

2. Emotional trust: The effect is mediated by emotional trust. Consumers feel less secure about relying on products explicitly labeled as AI-powered.

3. Risk perception: The negative effect is amplified for high-risk products (e.g., cars, medical diagnostics) compared to low-risk items (e.g., vacuum cleaners).

4. Consistent results: The effect held across various product categories, from TVs to customer service interactions.

It doesn’t help that companies are plying people with AI products lately, and some AI-driven campaigns are receiving serious backlash.

Most notably of late, Google had to pull its "Dear Sydney" Olympic ad after criticism of promoting AI-written fan letters. Apple and Toys "R" Us also faced pushback for pushing AI over human creativity.

What this all means for marketers:

1. Reconsider explicit AI labeling: The term "AI" might be trendy, but it could be hurting your sales if poorly placed or overused.

2. Focus on benefits: Emphasize product features and advantages rather than the underlying technology.

3. Build trust first: For high-risk products, prioritize establishing emotional trust before highlighting AI capabilities.

4. Context matters: The impact of AI terminology may vary based on your product category and target audience. Test messaging carefully.

The takeaway is a familiar one: while AI can be a powerful tool, how we use it and communicate about it matters.

Hope you enjoyed this week’s issue. If you missed it last week, you can read it here.

Until next time. Happy marketing.

—The AI Marketer