Has ChatGPT gotten worse recently? I had to convince it Biden was not currently president and that Kamala ran against Trump

Assessing Recent Changes in ChatGPT’s Performance: A User’s Perspective

As artificial intelligence tools like ChatGPT become increasingly integrated into daily workflows and decision-making processes, understanding their evolving capabilities and limitations is essential. Recently, some users have observed potential shifts in ChatGPT’s responsiveness and accuracy, prompting discussions about whether the model’s performance has diminished.

A User Experience with Political Data Interpretation

One user shared their experience starting with historical data inquiries about the years 2010, 2015, and earlier. During this process, they shifted the conversation to discuss current political figures, specifically Kamala Harris and Joe Biden. The AI initially provided correct information but then corrected the user’s statement regarding Kamala Harris’s election history, clarifying that she has not run against Trump as a presidential candidate, only serving as vice president.

The user then inquired about the current president in 2025. ChatGPT responded confidently that Joe Biden is the president, citing his victory over Trump in the 2020 election. However, the user pointed out that the 2024 presidential election is still pending, and thus, Biden’s presidency could not be ongoing as suggested. After persistent clarification, the AI eventually acknowledged this inconsistency, apologized, and corrected its statement.

Additional Observations and Concerns

Beyond this specific interaction, the user reported noticing other inaccuracies and a decline in the AI’s nuanced understanding. Notably, ChatGPT initially denied knowing about Kamala Harris running in the 2024 election, despite having checked current 2025 data. The explanation provided was that the AI was confused by the conversation revolving around historical political data, which the user argues contradicts the model’s supposed data cutoff date prior to 2020.

The user expressed concern that ChatGPT’s ability to handle complex or nuanced legal and political analysis has weakened significantly. They observe that the AI now seems more prone to contradictions and requires explicit prompts to reconcile conflicting statements, reducing its reliability for detailed analytical tasks.

Implications for Users and Future Expectations

This anecdotal account raises important questions about the evolution of AI language models like ChatGPT. As AI systems are updated and refined, users may experience shifts in performance—sometimes improvements, but potentially also regression. Transparency regarding model updates, training data, and known limitations remains crucial to maintain user trust.

For professionals relying on ChatGPT for legal, political, or analytical insights, skepticism about accuracy and consistency is warranted, especially given reports of recent declines in nuanced reasoning abilities. Ongoing feedback from users