DeepSik vs. ChatGPT: Similarity in 74% of Cases

A new study published by Forbes shows that the Chinese artificial intelligence DeepSik presents a similar writing style to ChatGPT in 74% of cases. These findings have attracted the attention of AI and natural language processing (NLP) experts, raising questions about the advancements of language models and their differences. The research, whose methodology has not yet been fully published, indicates that DeepSik has a significant similarity to ChatGPT in terms of writing style, word choice, and information organization. This result reflects remarkable progress in AI models' ability to imitate writing styles and understand human language. However, experts believe that similarity in writing style does not necessarily mean equal quality. The accuracy of information, diversity in responses, and the ability to provide creative content are also determining factors in assessing the quality of a language model. This similarity has also led some experts to accuse the creators of DeepSik of using ChatGPT's database for training this Chinese AI. DeepSik and ChatGPT have various similarities and differences, the most important of which are as follows: 1. Language models used: ChatGPT, a product of OpenAI, is trained on a transformer architecture and a vast dataset of text. DeepSik also utilizes a similar architecture, but it is unclear how the amount and type of its training data differ from ChatGPT. 2. Writing style: Both models have the ability to produce fluent and structured texts. The conducted study shows that DeepSik produced outputs in 74% of cases that were indistinguishable in style from ChatGPT. 3. Accuracy and reliability of information: One of the main challenges of language models is providing accurate and up-to-date information. ChatGPT benefits from a vast database, but DeepSik is still in the early stages of evaluation regarding information accuracy. This research holds significant importance in the development of language models. The high similarity between DeepSik and ChatGPT may indicate that other models are approaching the performance levels of pioneers in this field. This competition could also lead to improved model quality, increased information accuracy, and reduced potential biases in text generation. At the same time, these similarities raise questions about the authenticity and real differences between AI models. If different models are so similar, will real competition in this field continue? The new study has shown that DeepSik can provide outputs similar to ChatGPT in many cases, but there are still various factors that require closer examination. The future of language models is likely to trend towards increased accuracy, improved creativity, and reduced potential biases. This research is just a step towards better understanding the differences and similarities of advanced AI models, and certainly, more research will be needed for a better understanding of this field.