Back to Main
This Week in AI — 👀 Understanding how the world uses AI
AI

This Week in AI — 👀 Understanding how the world uses AI

September 16, 2025 | By: Dave Isaac

Each week we’ll gather headlines and tips to keep you current with how generative AI affects PR and the world at large. If you have ideas on how to improve the newsletter, let us know!

What You Should Know

 

OpenAI, Anthropic Show Big Comms Opportunities in How the World Uses AI

You often hear that AI models are black boxes, mysterious in how they produce results, and difficult to understand how they’re used. OpenAI and Anthropic pulled back the curtain on Monday by unveiling usage data that reveals two big opportunities for communicators. 

First is the growing use of AI for finding information and making purchasing decisions. OpenAI’s analysis found that nearly half of ChatGPT prompts fall into the category of “asking” for guidance, perspective, and advice, while 40% of usage is for “doing,” like creating content, planning, or coding. 

When ChatGPT came onto the scene, search wasn’t among the top use cases, but it’s surging now — AI referrals were up 357% year-over-year in June, and they’re showing different results than traditional search. 

While AI was initially billed as a productivity aid, it is increasingly a new entry point to information that influences how stories, brands, and even reputations surface online. This is a growing channel you now need to monitor for reputation and visibility.

Second is the opportunity to automate aspects of your communications work. Anthropic’s third economic index shows that while everyday users turn to AI for advice, companies using APIs increasingly treat AI as a worker. It’s the first time Anthropic found automation (49.1%) to be more common than augmentation (47%). 

More companies are putting trust in AI to quietly run parts of their workflow, and comms teams need to start thinking about which processes can safely be automated and which demand human oversight to preserve voice, nuance, and trust.

Many aren’t there yet, according to PRWeek and Boston University’s first AI in PR survey. It showed that employees were more likely to say their employers foster innovation and collaboration than that they use AI often in their work. Many also perceive their peers use AI more extensively than they do.

All this research shows that teams working to capitalize on AI search for visibility and automating parts of their workflow can get ahead. With clearer insight into how these tools are being used, communicators can move from reacting to AI to directing it. 

That means understanding how often your brand shows up in AI search, what the models say about your business, and what sources they cite. It means structuring content, FAQs, press materials, and online messaging so that brand-aligned answers rise to the top when prospects or journalists turn to AI for guidance. 

It also means deliberately choosing which parts of the workflow are better handled by automation, like drafting first-pass copy, building media lists, summarizing coverage, or handling routine reporting. Communicators exploring automation now can differentiate themselves and their departments and advance their careers. 

You can turn a black box into a tool you can shape to do better work and get stronger results.

Elsewhere …

‼️ Double vision

What’s happening: There are plenty of AI tells (we’ve listed lots of them!) that are fairly easy to spot, but repetition is less obvious when you’re reviewing AI-generated copy, yet just as pervasive as “delve” or “not just X, but also Y.”

What we mean: Repetition doesn’t just mean using the same word or phrase several times throughout content. AI has a propensity to make the same point over and over again, using differing language each time. It’s one of the things to look for if you get an AI output that’s far longer than you’d expect. It’s especially common among articles with subsections where a point in the introduction section is repeated just after the first subhead.

How to stop it: While AI isn’t always a great editor, it’s pretty good at catching the different kinds of repetition when you give a basic command, like “I want you to review this and point out areas that are redundant.” It will identify repeated words, phrases, and even themes. 

Of course, you’ll want to confirm this with the eye test, because it may say “I’ve found the word encounter seven times in here” when it’s really four, or perceive a statement to be repetitive when it’s really just tangential to another point you’ve made.

Quote of the Week

“We have this generational opportunity to reignite American scientific leadership and to renew America’s position as a leader in innovation around the globe.”

— Markham C. Erickson, Vice President of Government Affairs and Public Policy at Google, in remarks at WIRED’s AI Power Summit in New York on Monday

How Was This Newsletter?

😀 😐 🙁