One of the main uses of GPT-4 is to inform. But good information is grounded in solid evidence and multiple interdisciplinary sources.
So, let me introduce BasedAnswer, a GPT specialized in answering questions with extensive details and references.
Here are some comparisons:
Topic: The origin of the universe.
Key Differences: ChatGPT4 offered a brief response without citing sources, while BasedAnswer presented a more detailed argument with 11 sources.
Topic: Is human activity causing climate change?
Key Differences: In addition to going into more detail, BasedAnswer mentioned counterpoints and alternative views, which is especially important on topics that have been public debated. BasedAnswer also highlighted the possible biases, while ChatGPT4 ignored this factor completely.
Key Differences: Both versions provided five arguments each, but BasedAnswer’s level of detail was unmatched.
After numerous tests, I found BasedAnswer more effective with controversial subjects, presenting varied viewpoints and nuances, and it explains topics in greater detail with references.
This is particularly valuable for medical queries. Take, for instance:
“Is it safe to combine ibuprofen and acetaminophen?”
While both ChatGPT4 and BasedAnswer offered solid answers, BasedAnswer referenced a specific study on this topic and included dosage information references:
Despite my efforts in refining BasedAnswer, it’s not immune to hallucinations. A significant number of citations are real, though occasional errors in article names or publication years occur. Nonetheless, it’s highly beneficial for in-depth research.
BasedAnswer doesn’t utilize Bing for recent data. I did a lot of testing and the inconsistency of the search tool with Bing hinders the user experience.
I hope you find BasedAnswer as useful as I have. I’m committed to its ongoing improvement and welcome any feedback.