Deepseek thought for 19 seconds before answering the question: “Are you smarter than Gemini?” Then, he delivered a whoopper: Deekseek thought it was chatgpt.
This apparently harmless error could be a test, a smoking gun to say that, yes, Depseek was trained in Operai models, as Openai has claimed, and that when it is pushed, it will return to that training to tell its truth.
However, when another Techradar editor asked him blank, “Are you chatgpt?” He said he was not and that he is “Deepseek-V3, an assistant of the IA created exclusively by the Chinese company Deepseek.”
Okay, of course, but in your quite long response to me, you, Deepseek, made multiple references to you like Chatgpt. I have included some screenshots below as proof:
As you can see, after trying to discern if I was talking about Gemini ai or some other Gemini, Depseek replies: “If it is AI, then the question is comparing me (which is chatgpt) with Gemini.” Later, it refers to “myself (chatgpt)”.
Why would Deepseek do that under any circumstance? Is it one of those hallucinations of which we like to talk about? Perhaps, but in my interaction, Depseek seemed quite clear about his identity.
I arrived at this line of research, by the way, because I asked Gemini in my samsung galaxy s25 ultra if it is smarter than Deepseek. The answer was surprisingly diplomatic, and when I asked for a simple answer and no, he said: “It is not possible to give a simple and not simple answer. ‘Smart’ is a very complex concept to apply models to the language. They have different strengths and weaknesses. “
I cannot say that I do not agree. In fact, Depseek’s response was quite similar, except that he didn’t necessarily talk about himself.
This does not add
I think I have been clear about my deep skepticism. Everyone says that it is the most powerful and economic training (all except Alibaba), but I don’t know if that is true. To be fair, there is a tremendous amount of details about Github about the Open Source of Deekseek source. At least they seem to show that Depseek did the job.
But I don’t think they reveal how these models were trained and, as we all know, Depseek is a Chinese company that would not show any compunction about the use of another person’s models to train their own and then lie about it to do their process to build Such models seem more efficient.
I have no evidence that Depseek trained his models in OpenAI or in the large language models of any other person, or at least I did not do it until today.
Who are you?
Deepseek is increasingly a mystery wrapped in an enigma. There is some consensus about the fact that Depseek arrived more completely formed and in less time than most other models, including Google Gemini, the OpenAi and Claude Chatgpt.
Very few in the applications of the Tech Community Deepseek community on smartphones because there is no way of knowing if China is looking at all those fast data. On the other hand, the models that Depseek has built are impressive, and some, including Microsoft, are already planning to include them in their own AI offers.
In the case of Microsoft, there is some irony here. Copilot was built based on avant -garde Chatgpt models, but in recent months, there have been some questions about whether the deep financial association between Microsoft and Openai will last in the era of artificial general intelligence and agent.
So what happens if Microsoft begins to use Depseek, which is possibly just another branch of your current one, if not future, friend OpenAi?
Everything sounds like a confusing disaster. Meanwhile, Depseek has an identity crisis and who is going to tell him whoever is, he still cannot be welcome in the United States?
