Have you compared your perplexity and ChatGPT responses?

[Renegade]

In general I have noticed that Perplexity is giving more generic responses which are more like a refined search engine output. It is not really diving deeper into the context and history to give more tailored useful responses, leaving work to be done by the user.

Even with GPT 4.o model selected the quality of responses was generally inferior compared to what I am getting on ChatGPT native tool. It was rather surprising. Is it because GPT is building your profile in the backend and using that also as a context?

Not saying that perplexity is not good or useful, just lesser.

People who have used both, how has your experience been?

 

[Fenix]

I tried asking it the question about profiling me and ChatGPT said No.

 

[zr0]

Is it because GPT is building your profile in the backend and using that also as a context?

gpt seems to use your entire history as context in any isolated instance, which makes its responses feel more tailored – though it will deny doing so if you ask

perplexity's responses lack the personalized depth that comes from integrating previous conversations, it tends to stay more agnostic that way – they may seem suited for different applications, but i find myself going back to gpt every time i try perplexity way more than the other way around

 

[Fenix]

Ask GPT to delete all your chats, that way the response won't be tailored.

 

[Renegade]

but i find myself going back to gpt every time i try perplexity way more than the other way around

This is exactly it. I dont want to delete the chats and the responses to not be tailored.

Its nice that I dont have to play 50 first dates with the chatbot before I can get something meaningful done. So perplexity is not that great from that perspective.

 

[krish_puri]

I typically use Perplexity when I need to ensure that the information I receive is completely accurate and supported by reliable sources. In contrast, ChatGPT excels at content writing and creation, although it may sometimes fall short in terms of accuracy.

 

[Spout3628]

This is exactly it. I dont want to delete the chats and the responses to not be tailored.

Its nice that I dont have to play 50 first dates with the chatbot before I can get something meaningful done. So perplexity is not that great from that perspective.

have you tried perplexity's writing mode? It bypasses the extra steps that perplexity does

 

[Renegade]

What is this mode, what does it do and how do I turn it on/off?

 

[mayank11280]

I think, for my use case, Perplexity generates better responses. ChatGPT is good to create large content but Perplexity is good for research and knowledge gathering.

I have found Perplexity's source citation to be more accurate and helpful.

The idea of Spaces in Perplexity is also very helpful. We can tailor each space for particular use cases. I tend to isolate each space for different type of study. Lack of interconnect between different spaces helps maintain a degree of agnosticism in response, which I feel is necessary for unbiased answers.

Additionally, each space can be further customised with prompts and documents. Say, I upload one or more documents in one space and ask different questions in different context windows within that space about the document/s. It helps so much with rapid analysis of the document/s.