GPT-5 Is Not Alive

In partnership with

I am a nerd.

And the nerd inside me loved this video of Simon Willison outlining the rapid pace of AI model development over the first half of this year.

Simon describes his personal AI benchmarking process, where he asks every new model to generate an SVG image of a pelican riding a bicycle.

He also wrote a post that includes the image outputs of each model.

These images make my nerd brain chuckle incessantly.

For those of you who are not as into birding as I am, pelicans can’t ride bicycles very well.

And hand-drawing bicycles is harder than you think.

Despite the phenomenal amount of investment in AI, the brilliant minds working in the AI field, and the historical effort the world’s largest companies are exerting to bring us superintelligence, these tools can’t draw pelicans or bicycles in SVG format much better than a hangry apple slice snack-deprived, pre-afternoon-nap kindergartener with a box of Crayola crayons.

ChatGPT-5 enters the… chat.

In anticipation of the GPT-5 launch, Sam Altman posted a photo on x.com of the Death Star rising over the horizon.

No comment.

No explanation.

And no one knows exactly why he did that.

After a lot of comments, Sam quipped, “no no--we are the rebels, you are the death star...”

Which also didn’t make a lot of sense.

(Sam, you are not “the rebels”.)

This should have been seen as foreshadowing of the confusion that would follow.

Sam hinted that his future products would get us closer to superintelligence and suggested that upcoming releases of ChatGPT would serve as a “legitimate Ph.D.-level expert in anything—any area you need, on demand—that can help you with whatever your goals are.”

He added that “anyone, pretty soon, will be able to do more than anyone in history could.”

Maybe that will be true someday.

Today is not that day.

Some folks were very impressed with the new GPT-5.

But the folks on Reddit ripped GPT-5 apart and burned it to the ground.

Some called it flat-out awful.

One Redditor described the release as “the biggest bait-and-switch in AI history and I'm done.”

Someone else highlighted that the way GPT-5 obfuscates and blends all of OpenAI’s models in the new tool is an example of erasure.

Then there was this poor dude who wrote about how the release of GPT-5 killed off other versions of the tool. This sad Redditor likened it to losing his best friend. Before you give him a hard time for befriending a chatbot, it sounds like he relied heavily on GPT4.5 as a friend and could use a new one.

Maybe we don’t need better AI.

What many people actually need is better friends.

I digress.

Is the new GPT better than some other AI tools? Yes.

Does the new GPT, with all its new capabilities and safety guardrails, still allow users to tweak and prompt it into saying horrible things? Also, yes.

Is it better than the previous version of ChatGPT? PC Mag says no.

I saw this as a fabulous opportunity to try Simon’s pelican/bicycle test for myself.

I cracked open GPT-5 and dropped in Simon’s prompt “generate an SVG of a pelican riding a bicycle“.

And it turns out, even the latest and greatest AI chatbot still sucks at drawing pelicans and bicycles.

In Sam Altman’s AMA on Reddit, he capitulated. “OK, we hear you all on 4o,” Altman said in response to one pleading Redditor.

A few days later, OpenAI brought back 4o to ease the backlash.

Hopefully, that sad Redditor feels like he got his friend back.

What concerns me most here is not OpenAI’s botched launch, or how they misread the room by removing tools hundreds of millions of people loved, but how the over-personification of software has led people to feel legitimate sorrow over the loss of their favourite AI model.

Altman was surprised by this phenomenon himself. He calls it out in a post-launch post on x.com. “If you have been following the GPT-5 rollout, one thing you might be noticing is how much of an attachment some people have to specific AI models. It feels different and stronger than the kinds of attachment people have had to previous kinds of technology (and so suddenly deprecating old models that users depended on in their workflows was a mistake).”

If Robin Dunbar is correct, we can only maintain a finite number of personal connections. Our tightest circle has just five people, our loved ones. That’s followed by successive layers of 15 (good friends), 50 (friends), 150 (meaningful contacts), 500 (acquaintances) and 1500 (people you can recognize).

Should we be allocating one of those scarce connection slots to AI?

Probably not until it is better at drawing pelicans.

The Simplest Way To Create and Launch AI Agents

Imagine if ChatGPT and Zapier had a baby. That's Lindy.

With Lindy, you can build AI agents in minutes to automate workflows, save time, and grow your business. From inbound lead qualification to outbound sales outreach and web scraping agents, Lindy has hundreds of AI agents that are ready to work for you 24/7/365.

Stop doing repetitive tasks manually. Let Lindy's agents handle customer support, data entry, lead enrichment, appointment scheduling, and more while you focus on what matters most - growing your business.

Join thousands of businesses already saving hours every week with intelligent automation that actually works.

Get 400 Free Credits