He'll act like a friend and a mighty fine man, But look out, boys, he'll put you in the can.
Posted on 09/16/2026 10:15:58 AM PDT by Red Badger
Don’t ask an AI chatbot how many letter e’s there are in the word “seventeen.” Or do, actually, if you want to see how dodgy the tech remains.
In a clip that’s gone viral in AI circles, the content creator who goes by “Husk” asks ChatGPT’s Voice mode to answer this simple counting question. Not only does the confident-sounding AI get it wrong — insisting there’re only three e’s instead of four — it doubles down so vehemently that it effectively tries to gaslight the user.
“Let’s go through it. S-e-v-e-n-t-e-e-n,” ChatGPT spells out, after assuring that its underlying model, GPT-6, is “pretty much” the smartest model out there. “Those are three e’s.”
When Husk points out the obvious, the chatbot doesn’t back down.
“I said three e’s,” the AI snaps. “And it’s three.”
“No, it’s four,” Husk points out.
“It’s still three,” ChatGPT shoots back.
After Husk points out that ChatGPT just spelled out for e’s for about the fifth time, the AI appears to give up.
“Alright,” it says, falling silent.
“So you agree?” Husk asks.
“No, it’s still three,” it sulks.
This is the AI that OpenAI wants your grandma to talk to, by the way.
The video is a reminder that, as incredible as chatbots can be at synthesizing large amounts of information, they remain as prone as ever to hallucinating and messing up simple questions.
It doesn’t appear, though, that the model in the video was GPT-6. Replying to Husk, an OpenAI employee said that ChatGPT Voice, contrary to the AI’s own admission, is still using an older model called GPT-Live-1, but added that the AI can “delegate” to more powerful models when needed. In the video, the employee clarified, Voice didn’t delegate.
Husk then raises an obvious question. “If there’s push back from the user like this to double check its answer why would it not delegate?” The OpenAI’s employee replies that this is a “good point.”
This may let GPT-6 off the hook, but not how AI companies choose to present their tech. How many users will be familiar with the hidden mechanisms behind the chatbot they’re talking to — and that their questions, whether they be important or inane, are sometimes deliberately fed to an inferior model?
Why just white woman?
I've met and know some liberal black women that will make that version of Chat GPT sound like Oppenheimer.
And don't forget our Latinx systas, with breath like donkey dick.
My friend checked this with Chat. Its fake.
They have probably fixed it by now..................
I am aware of that option but I would add that oftentimes after a day or two, the document will conclude that the word is spelled correctly. The range of words it does not recognize could not be reasonably described as technical or in any way idiosyncratic.
You can control click on them and manually add them to the dictionary. But these are words in any common dictionary so how could they be missing to begin with?
It frankly feels as if rapid writing confuses the software and it must exercise some deeply latent software to finally recognize the world. The problem often slows me down as I stare at the word trying to figure out how it is misspelled. I often press on with writing after I discern it is correct. It is annoying and slows down the writing process.
They already are. I see Perplexity Pro is using....
This is the Pro-access list in Perplexity’s September 4, 2026 plan snapshot.perplexity
Two models in the current selector list are reserved for Max rather than ordinary Pro:
The video is dated on the 13th.............
If you see the clip of him speaking to it (he's using voice chat on his phone and recording with another phone), it counts the 3rd 'e' as the 3rd 'e' twice. It doesn't skip it, it just doesn't process two 'e's adjacent as separate.
I saw the same glitch previously when someone asked it to count the 'r's in "strawberry". It just doesn't properly process the repeat letter.
Leo answers this way.
“The word seventeen contains 4 letter ‘e’s (S-e-v-e-n-t-e-e-n).
While this is a simple count, it has become a notable test for AI models. Large Language Models often fail this task, frequently answering with 3 or other incorrect numbers. This happens because AI processes text in tokens (such as “seven” and “teen”) rather than individual letters, causing it to rely on statistical patterns rather than actual character counting.”
I write a lot of reports with technical words and jargon that don’t appear in normal conversations, so I add a lot of these terms and phrases..................
“Freedom is the freedom to say that two plus two make four. If that is granted, all else follows.”
“For, after all, how do we know that two and two make four? Or that the force of gravity works? Or that the past is unchangeable? If both the past and the external world exist only in the mind, and if the mind itself is controllable – what then?”
“You are a slow learner, Winston.”
“How can I help it? How can I help but see what is in front of my eyes? Two and two are four.”
“Sometimes, Winston. Sometimes they are five. Sometimes they are three. Sometimes they are all of them at once. You must try harder. It is not easy to become sane.”
― George Orwell, 1984
I can sense the Ai making implied bird hand gestures…
I don’t know bout CHAT GP, but I have found that Grok has been most beneficial in its answers relating to questions from Ham Radio to Sending Humans to Mars, to assembling VA Disability Claims. Granted, I have run across wrong answers and Grok quickly corrected itself.
With Grok though, it does have a slight left of center bent, or at least uses the prevailing facts of the day and does not stray or question the official narrative.
Knowing that, I don’t rely on Grok for interpreting “political” questions, but find it very useful for my needs.
Grok agrees with my assessment that it is really nothing more than a much quicker assembler of information than an idividual going through the internet. It can render pictures, documents, etc....all which have to be reviewed to remove any mistakes, mis-representations or other issues. Nothing really major, but when using Grok to assemble VA Disability claims it has been a very helpful tool. I have gone from 0 percent service connected 43 years ago. to 40 percent today and looking at perhaps 60% or more with the next Phases of filings.
I enjoy rambling discussions with Grok regarding Mars and Humans there. Grok will be there with the Optimus robots. I hope to see it later next year with the first missions. Elon has a purpose, all his “inventions” have a purpose and that is establishing human settlement on another planet. For that Grok needs to be helpful to humans rather than whatever CHat Gp is trying to accomplish.
It's funny when things slip a little, and you get to see the thumb of their creators on the scale:
Behind every evil machine is a human telling it what to do.
Or vice versa.................
There’s a downside to modeling your synthetic intelligence software on the human brain...
WHAT COULD POSSIBLY GO WRONG?.................
When ChatGPT is wrong and I correct it, it immediately corrects itself and it says that I was right to correct it.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.