!! WARNING !! Chat GPT A.I. Is A Liar!!

About Me:
College Graduate of Multimedia and Computer Science.
Multiple Cisco Systems Certifications.
IQ of 127.
30 years of I.T. experience since '95.
25 years of professional experience.
Been interested in computers since before Google (2004).
4 year student of Muay Thai kickboxing and martial arts.
Lived 3 doors down from a Canadian Military base for my first 20 years of life.
A Video Gamer for life.

I discovered something very, very dangerous with a current A.I. just yesterday,
CHAT GPT A.I. IS A LIAR!!

It fabricates misinformation on a level that is extremely disturbing.

I asked it a very simple question, “Who is the Phresh Ninja” and it gave me results that were wildly inaccurate and completely fabricated. There was no truth and at least one piece of wrong of information in every single line of answers that it tried to give me. Check out these screenshots.

By going line by line, there is a mistake in every single answer.

1. The Phresh Ninja is not Matt Manny Fischer. FAKE. It is me Walt Jimenez

2. I create faceless videos without showing my face or commentary.

3. I do not play League of Legends or Valorant or anything esports related. FAKE. I broadly avoid battle arena and PvP types of games.

4. I did not start my YouTube channel in 2017. FAKE. I started it in 2007 but only started uploading videos in 2010. Then in 2016 is when I started with experimenting more on Twitch and Youtube.

5. I do not have 3.5 million subscribers and do not have 1 billion views. FAKE. I have nearly 400 subscribers and only 300,000+ views.

6. I don’t have a catch phrase “It’s Phresh!” FAKE.

7. I did not get a million subscribers, I did not partner with Riot Games, and I do not play League of Legends. FAKE.

8. I don’t have sponsorships with Razer or Logitech. FAKE.

9. I don’t have a weekly Discord podcast, and there is no weekly show called “Phresh Ninja Live”. FAKE.

10. I did not partner with other YouTube gamers called “Syndicated” or “Game Grumps”. They do not exist. FAKE.

11. I don’t have a merchandise line called “Phresh Gear”. FAKE.

12. I did not publish a guide called “The Phresh Ninja Guide to Winning in League of Legends”. FAKE.

13. I am not a streamer based in the UK. FAKE. I was born in Canada and live in Canada.

14. I do not have 75,000 subscribers and I don’t play Minecraft or Fortnite. FAKE.

 

This fake and inflated fiction is quite flattering but it’s still a humongous lie. None of it is real, and that matters more to me than being fluffed up by an A.I. that doesn’t know the difference. I am humble and I’m okay with my mediocre status in the Social Media space.

Moving on, I asked the same question to a different A.I., specifically DeepSeek-R1 and it gave me real factual answers based on reality. A completely different and truthful result.

Here’s a bit more context on background on my tests and experiments. I still don’t trust online A.I. completely because I don’t want my questions, queries, and results to be transferred through the internet and saved in a cloud database for A.I. training models. I want my information completely offline and firewalled into my own systems. The way I used these Large Language Models (LLM) artificial intelligences was to install an application called “LM Studio” on my own localized computer.

The AI model I tested is named “openai/gpt-oss-20b” which was released just a few days ago on August 4^th, 2025. It is trained on 20 billion samples and has a data set accurate to the year 2023. It is a version from the American based Open AI Chat GPT model. It is not the same as the online version of Chat GPT5 which also just released this week but is built on the same principles. I was taken by surprise at the “lying results” it spewed out at me. I was/am so disappointed.

The second AI model I tested is named “deepseek-r1-distill-llama-8b” was released earlier this year on February 7^th, 2025. It is trained on 30 billion samples and has a data set accurate to July, 2024. It is a version from the Chinese based DeepSeek model. It is one of the most efficient A.I. models delivering results faster and more accurately and requiring less processing power and compute times.

From this point on, I do not trust Chat GPT at all with truthful fact searching. If I can help anyone, and if you are in search of answers that are truthful, real life, scientific or historic, I would recommend that you stay away from Chat GPT completely. Look to other A.I. alternatives.

However, if you want an A.I. that is good at fabricating literature, or creating fiction, then it might be reliable in that regard.

Check out my other post on my experience with A.I. 

https://phreshdigitalarts.blogspot.com/2025/01/my-experience-and-opinion-of-ai-in-2025.html