#309 – The State of AI: Keep Up Or Lose Out

🗒️ Overview

  • Get the lowdown on the results of surveying over 3000 marketers on AI
  • Find out if people really think AI is coming to steal our jobs
  • Learn how people are using AI

In case you missed it, we released the results of one of the largest AI surveys in the industry last week.

So today, we’re here to digest all the juicy bits of that survey and find out what you guys really think of AI.

Since you can go ahead and read the results in full at the link above, we’ll avoid repeating all those statistics in full here but instead focus on which insights stuck out the most.

What do marketers think about AI?

First up, some basic stats on AI usage for marketers:

  • Most marketers use it on a daily or weekly basis.
  • They consider it a highly important part of their job.
  • The number 1 tool used is ChatGPT, with Midjourney and Jasper.ai trailing behind.
  • Most marketers agree that a lack of knowledge about these tools is what’s holding back widespread AI adoption.

Frankly, it’s a no brainer: if you aren’t using AI and are a professional marketer, you’re probably going to fall behind.

The main problem, however, is that not many people are really tinkering or playing around to unlock its full power.

How to Use AI

With most people chasing the dream of 1-click content creation, many don’t realize there is a huge potential out there to use AI for multiple different purposes.

Mark has recently been using ChatGPT-4 to assist in writing outreach emails. Of course, it’s not a one-click process, but here’s his current workflow:

  1. Manually research the prospect thoroughly and collect notes
  2. He then writes a mini bio about himself
  3. He feeds ChatGPT a thorough, research-filled prompt and asks it to generate it for him.

The key here is that he doesn’t just put in a generic prompt and hope for the best – this prompt was backed up by a lot of manual research.

AI is, however, great at sorting all that data into a coherent output that isn’t just generic outreach spam.

On Gael’s end, he’s been using it to create highly detailed articles.

His process is a little more detailed, but here’s the gist of it:

  1. Start by manually researching competitors and which topics they cover
  2. Ask BingAI to create an outline on the same topic and combine the manual and AI-generated research together
  3. Ask BingAI to flesh out the master outline, finding additional examples and research notes
  4. Feed that outline into GPT-4 with a detailed editorial prompt custom-made for our site.
  5. Ask GPT-4 to edit its own article

By now, you can probably see these prompts go beyond your simple “Do X for me”.

A few other use cases Mark and Gael talk about include:

  • Subject lines
  • Social Media posts
  • Admin Tasks (e.g writing a letter of recommendation)
  • Brainstorming Ad angles
  • Writing workout routines
  • Fixing/Writing code for plugins

AI and Content Creation

Naturally, a lot of the respondents were most interested in AI content creation.

There were a few interesting insights based on the survey here:

  • Most people felt that AI could write about the same as your average human writer
  • Most people did also believe it had a lot of biases and inaccuracies
  • And unsurprisingly, as a result, most people use fact-checking in their AI content.

Perhaps the most shocking stat from this was that 11% of people openly admit to not fact-checking their AI-driven content. It may not seem like a lot, but scale that up to the entire internet, and that’s a whole lot of unverified information floating around the web.

Although, as Gael did point out, if you take AI out of the equation, just as many human-written articles, if not more, probably also go un-fact-checked, meaning that perhaps people who are against AI, for this reason, should reevaluate how they see this.

Detecting AI Content

And the final chunk of data Mark and Gael looked into revolved around the detection of AI content, disclosure, and how AI is used.

Again, here were the most insightful results:

  • Just over half the participants suspect they’ve received AI content from a paid writing service
  • The majority of people don’t bother with AI detection tools
  • Most people agree that companies should disclose if they’ve used AI to write content
  • And finally, most marketers agree that they would be happy to disclose that they’ve used AI in their content

There’s certainly a wider discussion to be had around whether it’s ethical to sell content without disclosing that AI helped. Gael believes we’re currently in a bit of a transition period.

We’re still paying human written rates for something which can ultimately be done faster and cheaper with AI – if handled correctly.

He anticipates that as “AI Operators” become more prominent (e.g, someone who doesn’t write but instead prompts AI to write), prices will actually go down. But since all of this is so new, we’re still in a transition phase to see that happen.

It’s also good to see people’s faith in AI detection tools in a suitable place. With OpenAI claiming their own AI detection tool only picks up 26% of AI written content, it’s pretty absurd to see schools using it to punish students!

So there’s a snippet of the results. If you’re curious to find more, you know where to head 😉