AI round-up: Week of March 24, 2025
Well, faithful readers (always wanted to say that), we have had one hell of a week.
I’m not going to take up much time here in the intro, but when you see image generation have its ChatGPT moment … you realize why Sam said the OpenAI GPUs are melting.
All the big guys had big news … but come on. It’s all about that new image gen tool.
Let’s get to it.
The Heavy Stuff
Story 1: OpenAI introduces a new image gen tool.
I would spend some time on the OpenAI page checking it out. But if you’re pressed for time, I played with it a bit.
First, I asked it to create a picture of a potential customer touring a manufacturing facility. The few specifics I gave were the group (five), the breakdown (one plant manager; four potential clients) and the purpose (touring for the intent of choosing them as a partner).
Then I asked it to create an image that depicted the group as concluding their tour and seeing it as successful, and have the plant manager take a selfie.
I didn’t like that the two guys on the right looked the same, so I asked it to drop two Akhia employees into the picture. One of them, a female, would be positioned as she was on the tour as the marketing lead and supporting the relationship side of the visit.
(See if you can guess which two employees I gave it. Answers at the end of this round-up.)
Seeing as the meeting was a success, I then asked it to create a direct mail kit that would be sent to the folks who attended the tour. I gave the touring company a name (MakeIt) and the manufacturer a name (Dynamix).
I then went and breathed into a paper bag.
Story 2: Gemini 2.5 is here.
And apparently, it’s balling out already. (TechRepublic)
Story 3: Microsoft Copilot releases two new AI agents. (ZDnet)
Meet Agent Research and Agent Analyst (maybe your newest co-workers?)
This is in addition to the six security agents Security Copilot released last week.
Story 4: DeepSeek gets in on the fun – v3 is here.
Story 5: AI-generated art does not qualify for copyright protection.
So … enjoy all the cool stuff you can make and do (stories 1-3), but guess what … you don’t own it! (SocialMediaToday)
Story 6: Remember bluelinks?
This WSJ article breaks down the reality of what’s to come. The title? “I Quit Google Search for AI – and I’m Not Going Back.”
I don’t think you can downplay this. As an agency, we should be talking to you about what this means (and with some of you, we have already). If you have any ownership over search, do not ignore this take.
Story 7: Thought we were done talking AGI? Guess again.
Paul Roetzer launched his podcast “The Road to AGI,” and I would highly, highly encourage a listen.
The not-so-heavy stuff
Story 1: Browser Use, the tool that fueled Manus, is gaining popularity and funding.
This might not age well in the ‘not-so-heavy stuff’ section, but time will tell…
Story 2: Google is rolling out Gemini’s real-time AI video features.
Let me rephrase that: Gemini can now see your screen and video feed to react/reply.
Story 3: ChatGPT has a new voice.
It’s more relaxed and calming. Because AI can FREAK PEOPLE OUT!!!
Story 4: You’re my boy, Claude!!!
Claude is the best at writing sincere emails that sound most like humans. Seriously, The Washington Post put this to the test, asking five LLMs to write an apology email. I love this experiment.
Story 5: The Creator Strikes Back.
Andor’s creator refuses to publish scripts so AI can’t train on them. (Andor Season 2 is coming on April 22!!!)
A few that don’t fit in either category
Story 1: Keep using your brain – even if AI can think for you.
One of the first things I said when ChatGPT broke on the scene was that AI would take some jobs. I’ve mentioned this before. What I may not have mentioned in a while is why I said that. I hated that I claimed this is why I believed that statement, but it’s because people are lazy.
They will view AI as an easy button and eventually will stop checking the source of where things come from (we arelady were well on our way to that reality); we will stop looking for ways we can enhance the work AI was returning (and why not – time is limited); and eventually, we will just stop thinking.
Had I written about that feeling, THIS is what it would’ve sounded like (although not as good as Celia Ford/Vox makes it sound).
Story 2: So, how are we feeling?
Turns out … excessive use of AI has some of us feeling lonely.
That’s what this study, done by OpenAI in partnership with the MIT Media Lab, revealed. The reason it’s here instead of in ‘the heavy stuff’? It’s incomplete, which is what I like about the study. It acknowledges it’s a little too soon to tell the impact … and we are yet to see how this AI tech will continue to expand. But it’s good to get a beat on it now so we have something to compare it to.
Story 3: How software engineers actually use AI (Wired)
This story (almost entirely visual) is SO interesting. I wish Wired would do one for marketers/communicators. Whatever field you’re in, you will want to see results like this for it.
Final note:
As of this writing, I am down to 1 team in the draft I did … and 3 in the one Claude did.
Here’s the one I drafted:
Here’s the one Claude drafted:
Answer to which employees were added to the OpenAI pic:
Me, apparently eating 8 Kalteen bars a day, and April.
-Ben
As a reminder, this is a round-up of the biggest stories, often hitting multiple newsletters I receive/review. The sources are many … which I’m happy to read on your behalf. Let me know if there’s one you’d like me to track or have questions about a topic you’re not seeing here.