Are LLM skeptics just incompetent or do they have different versions?

Title: Are Critics of Large Language Models Experiencing a Different Version, or Are They Simply Misunderstanding the Technology?

In recent months, I’ve observed a notable decline in the performance of my personal ChatGPT experience, particularly with the free, default version. This decline seems to be sporadic but is especially apparent among specific user groups or populations.

This observation has led me to question the underlying reasons behind the criticisms faced by large language models (LLMs). Are some detractors simply lacking the necessary understanding or skills to effectively utilize these advanced tools? Or could it be that they encountered their first interaction during a period when the system was experiencing an atypical low in performance?

The discrepancies in user experiences suggest that we might be dealing with different “versions” or states of the same technology. For me, LLMs have consistently demonstrated high utility, generating coherent and relevant content, with minimal hallucinations or inaccuracies. Conversely, critics often highlight issues such as hallucinations, factual inaccuracies, and perceived uselessness — claims that starkly contrast with my personal use case.

This divergence raises an important question: Are criticisms rooted in genuine limitations of the technology, or are they influenced by isolated experiences, misunderstandings, or temporary system issues? It’s possible that some criticisms stem from users who, perhaps unintentionally, interact with the models during suboptimal periods or lack the expertise to craft effective prompts.

I am curious whether others have observed similar phenomena—instances where their experience with LLMs contrasts sharply with public criticism. Understanding whether these differences are due to technical factors, user proficiency, or other variables is crucial for comprehensively assessing the true capabilities and limitations of large language models.

In sum, we should consider whether negative perceptions stem from actual systemic issues or from misinterpretations and variable user experiences. As the technology continues to evolve, fostering better understanding and effective usage may be key to bridging the gap between experience and expectation.