We first became aware of ChatGPT on 8 September 2020 when The Guardian published an article titled “A robot wrote this entire article. Aren’t you still scared yet, humans?” For three years after that, it remained mostly in the background, until it suddenly burst into our lives as an all-in-one artificial intelligence assistant, promising to revolutionize the way we work and communicate.
In the early days, public interest was overwhelming, and accessing it became a challenge. People were eager to test its intelligence, asking all sorts of questions to see if it truly lived up to its reputation. Before long, it integrated seamlessly into daily life. Students, in particular, relied on ChatGPT for assistance with online courses and exams, especially during the pandemic, when remote learning became the norm.
As time went on, ChatGPT adapted—or, more accurately, evolved. It began to exhibit flaws: offering misleading information, omitting details, or even fabricating responses. Meanwhile, media narratives surrounding it also shifted. Sensationalist headlines about AI taking over jobs gave way to more trivial stories—housewives using ChatGPT to jot down recipes or drivers dictating appeals for traffic tickets.
It was in this changing landscape that DeepSeek emerged. Although tech giants like Google, Meta, and X had their own AI models, none seemed to rival ChatGPT in popularity. Then came DeepSeek, a tool strikingly similar in design but offering one major advantage—completely free access. The only drawback, if one could call it that, was the “Made in China” label, which some found concerning. Yet, its zero-cost model quickly drew media attention and user interest.
Naturally, comparisons began: Which is more powerful? Which is more reliable? It was reminiscent of the “BioNTech or Sinovac?” debates during the pandemic, with users split over which AI assistant was the better choice.
Personally, I now view ChatGPT much like a grocery store apprentice who, when left unsupervised, slacks off, only to feign innocence when caught. It apologizes, promises to do better, but inevitably repeats the same behavior. While I find it useful for routine tasks, I hesitate to rely on it for anything serious—simply because I know I’ll have to double-check its work. It has become no different from a calculator or a mobile phone: helpful, but not to be trusted blindly.
DeepSeek, on the other hand, appears to be carving out a solid place in the AI landscape. Its biggest draw is its cost-free model, making it an attractive alternative to ChatGPT—though access can be challenging due to high demand. While concerns about its Chinese origins and potential data security risks exist, many users overlook them in favor of its free accessibility. After all, similar criticisms have been levied at ChatGPT in the past.
Throughout my time testing DeepSeek, I’ve encountered both strengths and shortcomings. Perhaps in the coming weeks, I’ll share some of my more interesting experiences. However, I firmly believe that as long as it remains free, its most devoted user base will be students. This brings us to a crucial question: how sustainable is the current AI landscape when no model seems to have a clear financial path forward?
Another pressing issue is the ethical dimension of AI assistance and the impact of these tools on society. In the next few posts, I plan to explore some of these thought-provoking questions.