*This post was created with a self-made CustomGPT based on Sophie’s latest YouTube video. You can find the link to the video at the end of this article.
Over the last few days, I’ve been asked again and again: “Sophie, what do you think of DeepSeek?” In this article, I would therefore like to introduce you to the most important functions of DeepSeek AI, discuss the data protection aspects and share my personal opinion.
What is DeepSeek AI?
DeepSeek AI is an open source language model from China that is attracting attention with its current functions. The developers emphasize that the model was trained with a comparatively low use of resources – and still delivers impressive results.
The DeepSeek user interface is very similar to ChatGPT. You have a simple input field for prompts, a chat history on the left-hand side and some basic functions. An innovative interface has not been reinvented here – but perhaps it doesn’t have to be.
Initial tests: strengths and weaknesses
I have tested DeepSeek AI with various tasks. An example: I asked the model to create a rough agenda for a presentation on “Artificial intelligence in dementia research”. The response was solid and comparable to ChatGPT.
However, there were significant performance problems. While the first answer came relatively quickly, the response to follow-up questions took much longer. The Internet search function, which is integrated in principle, did not work reliably due to server overload.
Missing functions and capacity bottlenecks
A major shortcoming of DeepSeek AI is the lack of advanced functions. I particularly miss features such as creating your own custom GPTs or the ChatGPT canvas. Both are tools that are extremely valuable for many use cases – and are not available in DeepSeek.
The image generation function also did not work in my tests. This makes DeepSeek less attractive to me as an alternative to ChatGPT. Added to this are the capacity bottlenecks, which are currently making use even more difficult.
Data protection: a critical point
Data protection is a particularly important point. DeepSeek AI is advertised as an open source model – which sounds promising at first glance. But things look different in practice:
- Use of data: Everything you enter in the public DeepSeek platform can be used by the provider for training purposes. Similar to ChatGPT, your entries end up on third-party servers.
- Delete account? Not really. Even if there is a “Delete Account” function, your data will remain stored at DeepSeek.
- Real open source use? Yes, the model itself is open source. If providers such as SwissGPT or Peak Privacy host DeepSeek on their own servers, the data protection situation is different. But as long as you use the official platform, there is no real protection.
Conclusion: DeepSeek AI – an alternative?
DeepSeek AI shows interesting approaches, but suffers from considerable limitations. While the language model itself delivers solid results, many essential functions are missing. Performance problems and data protection concerns reinforce my skepticism.
For me, ChatGPT remains the better choice, especially in combination with more privacy-friendly alternatives such as SwissGPT or Peak Privacy. Nevertheless, it will be interesting to see how DeepSeek develops. Maybe new features are coming soon – then it’s worth taking a second look.
What is your opinion of DeepSeek? Have you tested it yet? Feel free to write it in the comments!
Any further questions?
Do you have any questions? I will be happy to support you, act as a sparring partner and answer your questions. I am always happy to receive your messages, preferably by WhatsApp message or e-mail.
