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 Starts the AI Agent Brigade
By many accounts, this year will mark the dawn of agentic AI — tools that can perform tasks on their own. OpenAI kicked things off last week when it debuted a feature for scheduled tasks. The demos and examples were incredibly simple, basically amounting to the reminders feature on your smartphone, but ChatGPT can perform slightly more complex tasks than that.
One use case that communicators may appreciate (although it’s not novel) is suggesting ideas for social media or blog posts based on industry news. For example, you can set ChatGPT to scour the web for a particular topic at 9 a.m. every morning, create a roundup of news, and draft social or blog post ideas (or even entire drafts) based on the stories and your company’s perspective. It takes a more elaborate prompt and a good edit of the output, but it can give you a quick starting point for positioning your brand in the news cycle.
These tasks can be one-time or recurring, and can only be set via the web version of ChatGPT for now. To set up a task, use “GPT-4o with scheduled tasks” in the drop-down menu for models at the top of the screen and set up your tasks using natural language. ChatGPT will send a notification (see here for how you may need to adjust your browser’s notification settings) and an email when the task has been completed. You can manage all your tasks in one place.
To be clear, this is just a baby step toward agentic AI. OpenAI reportedly plans to launch a more sophisticated agent, code-named “Operator,” later this month. Surely other AI platforms will offer their own spin on AI agents, too. With more on the way, ChatGPT’s scheduled tasks are a good way to get your feet wet with agentic AI when the bigger products launch.
Elsewhere …
- Muck Rack Releases State of AI in PR Report
- Oscar Hopeful ‘The Brutalist’ Used AI During Production
- Trump Repeals Biden’s AI Oversight Order, Shifts Focus to Innovation-driven Policies
- ChatGPT’s Newest Feature Lets Users Assign it Traits Like ‘Chatty’ and ‘Gen Z’
- X Games Adds New Twist: An Artificial Intelligence Judge for Snowboard Event
Tips and Tricks
Are you offering enough context?
What’s happening: There has been plenty of digital ink spilled lambasting “AI slop,” the low-quality, generic writing that AI provides if you don’t give it enough context. Too many people on the internet simply copy and paste the output without refining it, which results in a mountain of vague, empty content. If you offer enough instruction and background information, you’ll get a better result, but it can be difficult to know what the AI needs.
Try this: So, ask it. Once you’ve entered your prompt and you think you’ve given plenty of context, ask the AI tool if it has any questions before it begins its assignment. Often, it will ask pointed questions that will help make the content better. For instance, if you’re working on a press release, it may ask you to elaborate on a target audience, key differentiators for the product or service you’re announcing, or the tone you want a quote to have.
However…: Sometimes the AI tool will ignore that step and go right to creating the content, so be sure to specify that it should provide an answer about what questions it has “BEFORE YOU BEGIN DRAFTING” (yes, typing that in all caps seems to help).
Quote of the Week
“I think 2025 is the year that we go from ChatGPT being this super smart thing that can answer any question you ask to ChatGPT doing things in the real world for you.”
— Kevin Weil, Chief Product Officer at Open AI, to Axios in an interview at the World Economic Forum
How Was This Newsletter?
