In the Loop: Week Ending 6/28/25
Last Week in AI: Apple, Jobs & Environment Last week, the scale of AI progress was matched only by the scale of questions it raised. Apple and Meta ma...
AI has a branding problem.
Yesterday, Perplexity, the small-but-mighty “answer engine” that’s the darling of many AI aficionados, released “Deep Research,” a new tool that can go look into a problem for you and come back with a research paper, fully annotated, based on information collected from across the web.
It would be a very big deal and truly newsworthy if not for the fact that OpenAI released a product just a couple weeks ago called “Deep Research” that does basically what Perplexity’s Deep Research does. Or that Google released an AI-driven research tool last December called, you guessed it, “Deep Research”.
The people who work at these companies must hang out together at the same Silicon Valley coffee shop…did none of them compare notes on their respective product names before launch?
This isn’t the only problem with the way AI goes to market.
ChatGPT’s product offerings are advancing so rapidly that they have a dropdown box that displays no fewer than seven models that a user can choose from. And the names are so strange that it’s nearly impossible to decide which one to use. “Is 03-mini good for what I’m working on, or is 03-mini-high better?”
Another sign that they're moving quickly? GPT-4, which was a marvel when it was released on March 14, 2023, is now simply known as “Legacy model”…ouch…irrelevance happens quickly in AI model land.
OpenAI isn't the only one causing model confusion. Google’s Gemini model has four options to choose from, and even though they do a better job describing the differences between them, it’s not clear what kind of use case would cause you to choose one vs. the other – or why they don’t just put them all into one model.
Sam Altman from OpenAI must have heard the plaintive cries from we, the AI acolytes, because this week he announced that OpenAI “hates the model picker too” and will soon be doing away with it. Hopefully the rest of the gang at the Silicon Valley coffee shop will get the memo.
So while Altman has given us hope that the model picker challenge will soon be a thing of the past, AI’s brand confusion issues don’t stop there. Take the word “copilot”. It’s used by many companies, including Microsoft, as a name for its products. It’s also used by companies like HubSpot as functionality embedded into its products. But that's not all. AI companies also use the word “copilot” as a general term meaning “a technology that rides alongside you and helps you out.” And don't get me started on the word "agent".
No wonder we're all confused.
And from a UX perspective, the color purple is everywhere, as are images that convey the idea of magic – there weren’t this many wands and sprays of stars in all the Harry Potter movies combined.
So you can be forgiven for being utterly confused about which AI tool to use, never mind which model to use. And it’s OK to be unsure if “copilot” should be capitalized or not. And completely fine to run screaming from all manner of sparkling purple software.
We all know that AI is advancing at lightning speed, but can just one of these companies hire someone who understands branding and go-to-market and give our poor addled brains a break?