What is the best AI ChatGPT or Copilot, Cursor, Grok?
Last Updated: 01.07.2025 06:38
The real threat isn’t smarter machines — it’s a dumber humanity looking for optimization, automation, efficiency and productivity at the human costs.
The “most accurate model to emerge from these tests, Perplexity from Perplexity AI, still answered 37% of its questions incorrectly. The village idiot award, meanwhile, goes to Elon Musk's chatbot Grok 3, which was wrong a staggering 94% of the time. Impressively bad.”
Will it ever do something that shakes us to the very foundation, shifts our world, forces us to think completely differently?
Transformer models like GPT are nothing more than over-optimized probability statistical data algorithms and software models.
So the question is: will AI ever be capable of representing, reflecting and reinventing REALITY?
An analysis of eight AI models, including OpenAI's ChatGPT search and Google's Gemini, by researchers at the Tow Center for Digital Journalism found that “they gave an incorrect answer to more than 60% of queries.”
They don’t understand content and context — they compute and calculate, interpolate and extrapolate it. A large language model (LLM) takes your prompt, smart or not, draws from trillions of tokens in its training corpus, and generates the most probable responses.
Or will it remain an endless remix of the past data, experiences, observations, theories and models.