The main advantage of using a chatbot like ChatGPT is to get quick and accurate answers to our questions. Instead of searching on Google and spending hours browsing various websites, we can ask Bing, ChatGPT, and other similar AI systems. Despite ChatGPT being considered the best among them, neither it nor Google Gemini have been free from making mistakes, such as generating images of “Nazis of other races”.
Google announced Gemini 1.0 in late 2023, being labeled as the most advanced AI model the company had. As you might imagine, Gemini competed against OpenAI’s GPT-4, which can be considered ChatGPT’s rival. The AI was segmented into various versions like Gemini Nano, Gemini Pro, and Gemini Ultra, differentiated by the complexity of the model and its multitasking abilities. It is a multimodal AI, capable of generating and combining information in formats like text, audio, images, video, and programming code.
Google Gemini makes mistakes when representing people from different periods
In the tests of Gemini vs. GPT-4, Google’s AI won in almost all cases, getting off to a great start. However, the AI encountered several blunders when asked to generate people from different periods. For example, when asked to create an image of the “Founding fathers of the United States,” the result included a man of color who did not resemble the original figures.
When asked to generate an American woman, Gemini showed red-haired, white, and women of color, illustrating significant diversity. When asked for US senators from the 1800s, the results included Native Americans, women of color, and an Asian man. Most critically, the AI generated images of Nazi-era German soldiers as people of color and Asians. A Google representative acknowledged that Gemini indeed produced inaccurate results when creating historical images and stated that they are working to resolve the issue.
ChatGPT does not escape mistakes and starts spouting nonsensical statements
Google Gemini is not the only AI experiencing inaccuracies; ChatGPT has also faced issues, mixing multiple languages and producing nonsensical statements. As shown by Sean Mcguire in X, ChatGPT spoke English but made errors and mixed in Spanish words and other languages. For example, it said, “cada type requires un bitta lámpara bajo punto,” later explaining it was merging Spanglish.
The advanced AI also amusingly responded with “Muchas gracias for your understanding.” At one point, ChatGPT kept repeating words without generating anything else, almost as if it was going through a rough mental state. Interestingly, both ChatGPT and Gemini encountered these issues on the same date, February 21, 2024.
This article is titled “The world’s greatest AIs, ChatGPT and Gemini, lose their minds on the same day – how is that possible?” and originally appeared on El Chapuzas Informático.
Need Help?
Can't Find What Your Looking For? Just fill out our form and we'll ask our trusted distributors