As you may have heard Google recently updated a help document called "Creating helpful, reliable, people-first content."

And if you never thought to check Google's help files for content guidance, well, good. I'm glad I'm not the only one.

I only found out because of an absolutely fantastic email from my friend Laura Jawad outlining what changed, and what you should do about it (and why). Read it here. And, if you're not on her list yet, you should be (do that here too).

There’s lots to process from the update but Laura’s email mentioned one thing that got my gears going. And that was the update to another guide from Google that covers using generative AI content on your website.

There really isn't anything surprising or controversial in it, but there was one thing in particular that I wanted to bring your attention to.

Fact-checking.

We all know that AI can be prone to errors. But it’s so easy to forget it because the AI is so confident in its answers. It's really important for us to remember that the AI is not thinking. It's not retrieving facts or dreaming up new ideas. It's just, as Google puts it, predicting a likely sequence of words based on its training data.

And it does it super fast!

Therein lies the problem. AI lets you get a lot more done a lot faster.

But since it’s error-prone, it's generating more review work than you've had to do in the past; although there’s even odds that work hasn't been assigned to anyone. 👀

Google recommends that all AI-assisted content be manually fact-checked before publishing (including meta descriptions and alt text). But it's not just AI-assisted copy you need to be reviewing thoroughly.

If you're using AI to:

• generate new skills for it to use,
• write instructions for other agents,
• create apps,
• analyze data,
• do anything at all…

You need to be reviewing its work. And that takes time, and a process.

If you’re a solopreneur all that responsibility falls to you. And if you have a team, it still falls to you — unless you set up a process to make sure it doesn’t.

For everything AI helps you or someone on your team create, you need to decide who the output gets sent to for reviewing, and error-checking. If you want to make sure it's getting done, have them document what they checked, what they found, and what they changed.

So, be honest. How diligent have you been with checking the AI’s work?

Me, when I first started using AI. I never even thought to check anything other than the copy it created. Now, someone reviews everything before anything gets shipped, or goes live.

That someone is still me while we work on training the team. *sigh* Don't cry for me, it's only temporary.

xo,
Sandra

​
​