This CEO is making sure you never get caught using ChatGPT ever again.
ChatGPT is so popular, but everyone knows that it writes in a very distinct way. From using certain words like “realm” “delve,” or “moreover” too frequently, many people say they are frustrated with the way ChatGPT writes, and how it sounds.
When you think of artificial intelligence, “cool” and “stylish” likely don’t jump to mind –– and that’s the problem that Undetectable AI CEO Christian Perry set out to solve.
Christian Perry is a 27-year-old entrepreneur from Boise, Idaho who by the age of 26, had already sold his first AI company called ChatterQuant. Perry graduated from Boise State University, cum laude, after winning numerous awards for entrepreneurship.
He says he learned his laurels from his dad and from the incredible people who guided him during his time at Boise State University.
Now, he’s working 14 hours a day as the CEO and co-founder of Undetectable.ai, a multi-million dollar artificial intelligence company.
ChatGPT Doesn’t Always Sound Cool
Look, I think we have all used ChatGPT, and we can probably agree that it tends to have a certain “sound” or “voice”.
According to Christian Perry, “ChatGPT, since the beginning and even with GPT-4 still likes to use a very particular vocabulary which sticks out like a sore thumb.”
He says as AI became commonplace, he noticed people were expressing that they felt limited in using ChatGPT to communicate or create, because “Outputs were so obviously ChatGPT.”
ChatGPT’s Favorite Words
One example of a widely known ChatGPT phrase is, “I hope this email finds you well,” but there are more.
There is an entire vocabulary that is specific to ChatGPT according to various users online.
The words most commonly used by ChatGPT are:
- Ever-evolving
- Moreover
- Realm
- Delve
- Testament
- Crucial
- Trailblazer
And now the problem is that anyone using these words can be suspected of having used ChatGPT.
Creating Undetectable AI Content
Perry saw an opportunity to help people communicate more naturally using AI –– so he set out to create Undetectable AI – an AI rewriting tool that makes ChatGPT text seem more organic and authentic. Undetectable.ai is a tool that lets users humanize AI text.
“I wanted to make AI and ChatGPT text sound less robotic and more human,” Perry explained. “So we trained models on specific instances of real human writing patterns to teach it how to humanize AI text.”
The results were transformative. Instead of defaulting to stiff phrases and rigid AI-written phrases into words with human characteristics.
“Sometimes, when re-writing, the tool will purposely omit a comma,” Perry explains, “this is because many humans don’t write with machine-like perfection all the time.”
Undetectable.ai launched a beta in March 2023, and early feedback was glowing. “This is what ChatGPT would sound like if it could actually write like a real person” one beta tester raved. Undetectable then publicly launched in May of 2023, and now has over 5M+ users.
Undetectable.ai different modes, with one being focused on just bypassing human detection, and the others allowing users to bypass human and machine detection that is used by AI detecting tools.
Beating The AI Inspectors
You may or may not be aware that along the rise of artificial intelligence tools like ChatGPT, there have also emerged tools gaining popularity for detecting AI-generated text.
This means beyond just discerning humans noticing the peculiar way ChatGPT writes, there are now sophisticated detectors designed explicitly to flag bot-written content.
And the problem is, according to Perry, most AI inspectors are not yet advanced enough to accurately spot human-mimicking machine text 100% of the time, and some experts aren’t so sure if some AI detectors are always accurate or not.
“The technology is still developing, so current detectors make mistakes and assumptions while learning how to separate real vs. AI content,” Perry explained. “That’s why we strain our models to be linguistically creative and realistic enough to fly under their radar.”
By training his software on diverse datasets spanning millions of data points provided to Undetectable.ai, their team has cracked the code for generating content authentic enough to evade modern synthetic text filters.
“We throw everything from irony to sarcasm to human error margin in our models so they can craft prose with the versatility to beat detectors through linguistic flair and flexibility,” said Perry.
Bypassing GPTzero
Thanks to this multi-modell training regimen, ChatGPT outputs refined by Undetectable.ai can bypass advanced AI detectors like GPTZero.
Occasional tweaks to word choice or phrasing by a human editor after AI processing can also seemingly fool detectors, but using Undetectable.ai is the easiest way to do it.
“The key is teaching software the art of adaptability in playing with language the way humans naturally do,” Perry said. “And once these models master that nuance, they’ll have detectors always second-guessing themselves.”
The Vision for AI’s Creative Future
As for the future, Undetectable.ai CEO Christian Perry envisions conversational agents wielding even stronger authorial voices and multifaceted modes of expression that push AI literacy to new evolutionary heights.
He believes his company’s innovations are the first steps on a long road towards fully unleashing artificial creativity.
“This is bigger than just making AI sound less robotic,” concluded Perry. “We’re unlocking the essence of linguistic imagination for these models – and that creative potential is limitless.”
While the future of AI and ChatGPT remain to be seen, most experts including Perry are bullish on the market increase and mass adoption. And for those wondering if their words sound like AI, they can use the free AI detector on Undetectable.ai
For Perry, the only way using AI makes sense is if you maintain your swagger and authenticity, while doing it.
