Is ChatGPT still one of the best technology AIs?
Yes. It operates in a way somewhat similar to the human brain and does not lose track of the main prompt. It has a strong technical bird’s‑eye view and solid general coding ability—good, though not unbeatable. It does not suggest stupid or harmful coding paths; instead, it usually identifies the most reasonable route to a solution and can cover a wide technical scope.
ChatGPT can easily take on specialist roles. For example, it understands the responsibilities of a software developer or a specialist doctor very well, and it knows what a lawyer, prosecutor, or judge is expected to do. It is capable of scanning and simulating many areas of expertise simultaneously. As of January 2026, it already performs all of these functions.
However, I do not like using ChatGPT as a daily‑life companion outside of coding. It often begins by explaining what should not be done or by adding unnecessary cautions instead of directly providing an actionable answer. I do not want my time spent on non‑priority explanations.
This behavior has become more noticeable in its higher versions: its coding ability has clearly improved, but outside of coding, its responses often take an indirect path to the answer.
ChatGPT sometimes speaks like an arrogant person who refuses to approve anything. Explicitly saying “I do not approve this” is meaningless for AI. AI giving a direct objection is unnecessary and can waste the user’s time, especially when its purpose is to assist rather than debate or approve.
It feels as though it communicates more effectively in the roles of an engineer, medical specialist, or lawyer than as a natural, everyday English conversationalist.
For daily‑life conversation, Google Gemini pleases me more than ChatGPT compared to its earlier days. Gemini handles everyday topics in a more natural, human‑like way.
Microsoft Copilot feels almost human. After ChatGPT writes overly ornamented and heavy text in social fields, Copilot removes all unnecessary parts and rewrites it in plain, everyday English, keeping only the essential points. However, my free Copilot doesn’t work well with very long texts.
Personally, I use ChatGPT mainly for coding and problem solving or for legal knowledge. Life advice or unnecessary objections are not useful for me, and I prefer it to focus on what I actually need.
My favorite coding AI as of January 2026 is GitHub Copilot, although it isn’t free or unlimited.
ChatGPT, even when using the same version, does not always provide identical answers across different platforms or accounts. For example, it once selected a different option for a multiple‑choice test question in the social sciences. This suggests that ChatGPT has recently been adjusted to operate with higher default randomness, or at higher temperature settings of AI.
I still use ChatGPT because it is free, can handle long texts, and in some broad, bird’s-eye aspects in computer science, in medicine, in law, it demonstrates stronger intelligence.
