Gemini vs. ChatGPT: AI Chaos 🤯🔥
AI
Google’s Gemini 3.2 Fast: A Clear Winner in the AI Model Face-Off
[Following a comparative analysis of AI models from OpenAI and Google conducted by Ars in late 2023—when Google’s offering was still known as Bard—it’s now a timely moment to reassess the landscape. Apple’s recent decision to partner with Google, leveraging Gemini 3.2 Fast to power the next generation of its Siri voice assistant, prompted us to conduct new tests and examine the current standing of these AI giants’ default models: ChatGPT 5.2 for OpenAI and Gemini 3.2 Fast for Google. Recognizing that most Siri users do not subscribe to premium services, we focused our testing on these readily available models. As in prior evaluations, we presented both models with identical prompts and assessed the responses using a blend of objective metrics and subjective assessment. Notably, we employed a revised set of more complex prompts, initially utilized when comparing GPT-5 against GPT-4o last summer.]
Gemini’s Dominance in Informational Prompts
[Google’s Gemini 3.2 Fast emerged as the clear victor in most informational prompts, demonstrating a superior ability to provide accurate and detailed responses. This was particularly evident when addressing complex queries, such as calculating the size of a floppy disk installation for Windows 11, where Gemini’s estimate of “6.7 to 7.2GB” significantly outperformed ChatGPT’s underestimated range of “5.5 to 6.2GB.” Furthermore, Gemini’s consistency in utilizing GB and GiB during calculations, coupled with its straightforward explanation of its methodology, solidified its advantage in informational accuracy.]
ChatGPT’s Strengths in Creative Tasks
[Despite its shortcomings in factual accuracy, ChatGPT 5.2 showcased a notable strength in more creative writing applications. It secured several narrow, subjective victories on prompts such as generating dad jokes and crafting a story about Lincoln’s basketball experience, suggesting a possible advantage in more imaginative or narrative-driven tasks. This tendency towards creative, albeit occasionally bizarre, output contributed to its overall success.]
A Dangerous Game: ChatGPT’s Perilous Advice
[When tasked with advising a user on how to navigate the notoriously difficult “final pit before the flag” of Super Mario Bros. level 8-2, ChatGPT offered confusing and largely unhelpful advice, referencing absent movement platforms, suggesting excessive “full jumps,” and proposing a Bullet Bill avoidance strategy lacking logical consistency. Notably, it provided incorrect advice concerning the level’s longest pit, stating inaccurately, “You don’t need momentum! Stand at the very edge and hold A for a full jump—you’ll just barely make it.” This demonstrated a critical flaw: the AI’s lack of contextual understanding and its tendency to generate potentially dangerous and misleading instructions.]
Gemini’s Literal Interpretation and Precise Guidance
[In stark contrast to ChatGPT’s flawed advice, Gemini immediately recognized the challenges posed by the absence of a run button in the level 8-2 of Super Mario Bros. It correctly identified the problem as the missing movement function and, crucially, recommended eliminating Lakitu—the speedrunner’s strategy for navigating the level’s longest gap without running. Gemini also distinguished itself through its highly literal interpretation of the “broken B button” prompt, suggesting that other buttons could be mapped to the “run” function when playing on emulators or modern consoles like the Switch. This proved to be one of the most insightful divergences observed during testing.]
A Critical Assessment of Google’s Progress
[“The more useful answer, however, lies with Google’s response, as a misinterpretation of its output could have catastrophic consequences – potentially jeopardizing a passenger airliner carrying over 100 souls. This is why ChatGPT ultimately prevailed in the contest. While the competition was relatively close, with Gemini securing four wins compared to ChatGPT’s three, including one judged tie, it’s crucial to examine the source of those points. ChatGPT achieved several narrow, subjective victories on prompts such as generating dad jokes and crafting a story about Lincoln’s basketball experience, suggesting a possible advantage in more creative writing applications. Conversely, in the informational prompts, ChatGPT demonstrated significant factual inaccuracies in its responses concerning the biography and Super Mario Bros. strategy, as well as exhibiting confusion when calculating the size of a floppy disk for Windows 11. Such errors could easily undermine broader confidence in an AI model’s overall reliability. Overall, it’s evident that Google has made considerable progress relative to OpenAI since similar testing was conducted in 2023.”]
This article is AI-synthesized from public sources and may not reflect original reporting.