The trillion dollar AI face-off💲

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: BIG announcements from OpenAI and Google

2. The Stack: What’s the longest article we can write with AI?

3. Consultant’s Corner: Google AI search goes global: here’s what we can do

Onwards!

AI MARKETING UPDATE

OpenAI and Google go head-to-head in generative AI face-off

OpenAI and Google have been one-upping each other in the AI race, with every announcement seeming to top the hype of the last.

This week, OpenAI fired the first shot with GPT-4o, a turbocharged version of their already impressive GPT-4.

GPT-4o is insanely fast and will be free with limits (Pro users still get priority).

It’s not quite as proficient as GPT-4 in some ‘hard tasks,’ though, including coding.

The headlines are that GPT-4o can handle real-time audio speech recognition and analysis, recognize emotions through visual cues, and offer exceptional multi-lingual support for text and speech.

Some of the demos were phenomenal.

Some marketing-relevant use cases that have popped up already involve taking meeting notes, brainstorming real-time research and insights, and handling customer inquiries in multiple languages.

You can also simply describe ideas out loud and create content hands-free.

I used GPT-4o to write a test article on AI marketing, complete with statistics and a case study, without touching my laptop.

I can head out on a walk, chat with ChatGPT on my phone, and create drafts ready to edit and refine once I’m back at my desk. Better than staring at the screen and typing.

This was possible before GPT-4o, but it brings it to the next level and makes these kinds of workflows way more tractable.

Google hits back

Google wasn't about to let OpenAI steal all the limelight. At their I/O conference, they unveiled some heavy-hitters:

1. Imagen 3: A souped-up text-to-image model that generates high-quality visuals with fewer artifacts and glitches.

2. Astra: A real-time, multimodal AI assistant that sees and hears.

3. Veo: Google's answer to OpenAI's Sora, Veo is a video generation model that promises to create stunning, high-quality clips from simple text prompts.

4. AI Teammate: An assistant that'll be integrated into Google Workspace. It can build a searchable database of your work emails, messages, and documents, making it a breeze to find important info when you need it.

Perhaps the biggest Google announcement for marketers, however, is its retooled version of generative search and AI Overviews, which will be rolling out globally across the year.

AI search engines generate text responses to queries, potentially shuffling real websites down the SERPs in the process.

So, let’s not speculate anymore. AI search is REALLY coming now. Plus, OpenAI and Microsoft are bound to hit back with their alternative in the next few months.

I guarantee that search engines will look, feel, and behave differently by the end of the year — that’s obviously massive for marketers.

We need to take preemptive measures and get our ducks in a row in preparation.

I explore this in greater detail in Consultant’s Corner below.

THE STACK

Endless AI-generated articles

Previous editions of this newsletter discussed how changes to the Google algorithm and search behaviors suit shorter-form articles (500 to 800 words) driven by subjective experience and authentic knowledge. Read up about that here.

But that doesn’t mean we never need to write longer documents like eBooks, white papers, etc.

While creating these longer pieces can be incredibly time-consuming, we can at least leverage AI to assist in the process. However, in most cases, AI outputs are limited, forcing us to stitch bits and pieces together into a coherent draft.

Today, with a couple of tips and tricks, we manipulate standard AI tools into writing 3k, 4k, or even 5k pieces. I call this the “double it technique.”

Here’s how it works:

First, you want to request your article with a long word count, e.g., 2000+ words. From my tests, this works best with Claude 3 Opus. GPT-4o seems to stall at longer wordcounts.

I used the simple prompt, “I want you to write a 2000+ word article on AI marketing. Please make sure it's 2000 words or more.”

You’ll likely see Claude 3 Opus lie about its output’s word count. Now, you want to find the correct word count, copy that in, and simply tell the model to “double it.”

You can repeat this process several times to write vast but coherent articles of 6,000 words or more.

My finished piece was around 6,408 words.

Unlike producing multiple pieces and stitching them together, this technique doesn’t just bolt new content to the end — it expands existing content, too.

You set tone of voice guidelines or other specifications in the initial prompt, and Claude seems to follow them perfectly throughout the article, so long as you remind it to.

Is the actual content any good? Actually, yes, it’s quite impressive. But like any other AI-generated content, it’s only a starting point.

This is an excellent technique for drafting long standalone articles or content that can be edited or split into smaller pieces for a blog series, courses, personal educational purposes, etc.

Really useful for gathering tons of research into a single, coherent, editable asset.

CONSULTANT’S CORNER

On Tuesday, Google announced a revamped AI search that will soon roll out to users globally. SGE is about to get real.

By the end of the year, AI summaries will become part of Google search results for over one billion people.

The marketing and advertising industries are already bracing for impact.

When people start researching a new topic or product, they often begin with simple, broad questions. For example, if someone's interested in getting into photography, they might search for "What is a mirrorless camera?" or "What's the best camera for travel?"

The problem is, a lot of companies have built their entire SEO strategy around targeting these kinds of basic, high-level queries. But now, with the rise of AI-powered search, those generic searches might not even lead to their websites anymore.

In fact, Gartner predicts that by 2028, brands could see a 50% drop (or more) in organic search traffic as users start to favor these new AI search tools. And if Google rolls out its AI features to everyone sooner rather than later, that number could be even higher.

Some are even predicting Google's AI move could cost web publishers around $2 billion. Yikes!

Basically, if your SEO is all about answering simple queries with simple content, you might be in trouble. It's time to start thinking beyond the basics and figure out how to provide value that the smartest AI can't match.

Keep calm and carry on…

It's time to start preparing for AI search. Here are four key tips you can action now:

• Start investing more in anti-fragile traffic sources like social media and email lists.

• Sources, quotes, and stats seem to boost AI search visibility, so use them liberally (this also plays to the strengths of Google’s recent algo update).

• Front-load useful content further up the page with bullet point summaries or ‘key takeaways.’

• Titles and feature images really matter for capturing clicks from the AI search section.

Explore more about AI search and how to capture more traffic as it rolls out in this previous edition.

As if Google’s March algo update wasn’t controversial enough…

But, hey, we're a resilient bunch. Let's stay ahead of the curve and deliver awesome value to our audiences.

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