Yeah, still even when he didn't do that, and allowed Perplexity AI to search the web, it appeared to do as good a job rewriting the text as ChatGPT, just the references were rubbish. So Perplexity should have won by 1 point in my opinion. but I like the video overall.
🎯 Key points for quick navigation: 00:00:00 *🤖 Introduction and Purpose* - The speaker introduces the video and compares ChatGPT with Perplexity AI with the intent to resolve the debate over which is better. - The speaker outlines tasks in research activities including literature search, paper explanation, data analysis, writing, and editing. 00:02:18 *📚 Literature Search Comparison* - ChatGPT and Perplexity are compared in identifying peer-reviewed papers. - Perplexity delivers more sources and sections, making it more effective in this task. 00:03:13 *💬 Paper Explanation Ability* - ChatGPT uses bullet points for explanations, focusing on section-by-section breakdown. - Perplexity pulls essential points and offers contextual insights, providing a richer explanation. 00:05:19 *📈 Data Analysis and Graphing* - ChatGPT successfully processes and graphs data, demonstrating proficiency in data analysis tasks. - Perplexity fails to deliver a graph, indicating limitations in this area. 00:07:48 *✍️ Writing and Editing Evaluation* - ChatGPT refines text for peer review but lacks contextual references. - Perplexity includes references but errors undercut its effectiveness. 00:10:05 *⚖️ Overall Conclusion* - Each tool has strengths; ChatGPT excels in data analysis while Perplexity is better for literature searching. - The decision depends on specific needs, suggesting a combined use of both platforms. Made with HARPA AI
This is true, however what's holding me back as that I have a few pins on some very useful GPTs (agents) how do you get around this using perplextiy? 🤔
The bullet points are fine if you just want to get a quick overview. One thing that would be nice is if they had an "expand" button by those bullet points. You can do it manually by copying and pasting it into the prompt though and say "expand on this".
@@strategictechnologist True. ChatGPT is good at that. I've learned in Gemini and Meta AI you DO need to copy and paste - which is sooooooo annoying! But the tradeoff to get up-to-date information is worth it to me with some things.
Dear Andy, thank you for the video and all the work you do. It is super helpful- both technically and psychologicaly. A little tip: the LLMs are still developed at a high pace. Therefore it would be helpful to define the version of the model you are using to do the test. For example say chatGPT 4o, or Claude 3.5. All the best!
I am SOOoooo LOVING this review!! I haven’t even finished watching it, but I paused to post this comment while ‘in the moment.’ I’ve been wondering about ChatGPT vs Perplexity for quite some time. I have paid versions of both and go back and forth between them depending on what I’m doing (I’m a health technology implementation scientist). Andy is being his usual hilarious self. I find myself chuckling through many of his videos, and I’ve watched A LOT of them. Andy, if you happen to see this comment, thank you so much for putting out such rigorous, provocative, instructive, and humorous videos. I can’t thank you enough for doing what you do.
#Objection This is an unfair comparison: ChatGPT is an LLM, Perplexity in its advanced setting includes several LLMs including the ChatGPT4o itself (the same one). If you have set Perplexity with GTP4o you are comparing the same LLM, but if you set it to Academics, the results will give you papers in depth! To access the advanced settings, you must clic your own profile picture!
You didn’t mention anything about the Perplexity “Pages” functionality or the “Playground” or “Collections”. You have more homework to do before a fair comparison can be made.
I'm already coming from your previous video, the one you recommended to watch at the end.. To be more accurate this one over here: ruclips.net/video/Ne-j_U2Vkfw/видео.html&ab_channel=AndyStapleton - So there you said that basically once I pay to Preplexity I can switch between LLMs from the settings. So right after watching this debate video it left me in many questions.. First and foremost, I really wonder which LMM did you pick for making this research debate.. ? If ChatGPT 4.0 model was selected wouldn't it give the exact answers as the ChatGPT 4.0 Plus (web browser) would give.. ? If you did set another LMM model for Perplexity (like Gemini or Llama 3 or else) and compared the results to what ChatGPT 4.0 Plus does, then you left me wondering why you did not mention it in the very beginning.. ? I've appreciated your videos because I thought they are transparent and some what a true reflection of a real PhD student experiences but I guess it comes down to having more views and all that..
Hi Andy. If you tell perplexity to add peer reviewed references as I text citations, it does perfectly. I use this regularly. 😊
In perplexity, When you attach a document and want to get information only from that document, set the focus to writing. So it won't search the web.
Yeah, still even when he didn't do that, and allowed Perplexity AI to search the web, it appeared to do as good a job rewriting the text as ChatGPT, just the references were rubbish. So Perplexity should have won by 1 point in my opinion. but I like the video overall.
🎯 Key points for quick navigation:
00:00:00 *🤖 Introduction and Purpose*
- The speaker introduces the video and compares ChatGPT with Perplexity AI with the intent to resolve the debate over which is better.
- The speaker outlines tasks in research activities including literature search, paper explanation, data analysis, writing, and editing.
00:02:18 *📚 Literature Search Comparison*
- ChatGPT and Perplexity are compared in identifying peer-reviewed papers.
- Perplexity delivers more sources and sections, making it more effective in this task.
00:03:13 *💬 Paper Explanation Ability*
- ChatGPT uses bullet points for explanations, focusing on section-by-section breakdown.
- Perplexity pulls essential points and offers contextual insights, providing a richer explanation.
00:05:19 *📈 Data Analysis and Graphing*
- ChatGPT successfully processes and graphs data, demonstrating proficiency in data analysis tasks.
- Perplexity fails to deliver a graph, indicating limitations in this area.
00:07:48 *✍️ Writing and Editing Evaluation*
- ChatGPT refines text for peer review but lacks contextual references.
- Perplexity includes references but errors undercut its effectiveness.
00:10:05 *⚖️ Overall Conclusion*
- Each tool has strengths; ChatGPT excels in data analysis while Perplexity is better for literature searching.
- The decision depends on specific needs, suggesting a combined use of both platforms.
Made with HARPA AI
Well, you can set Perplexity to access any LLM you want, including ChatGTP.
I have pointed out that.
This is true, however what's holding me back as that I have a few pins on some very useful GPTs (agents) how do you get around this using perplextiy? 🤔
The bullet points are fine if you just want to get a quick overview. One thing that would be nice is if they had an "expand" button by those bullet points. You can do it manually by copying and pasting it into the prompt though and say "expand on this".
Just tell it to expand on what it wrote. No need to copy and paste.
@@strategictechnologist I prefer to copy and paste to ensure it knows exactly what I want expanded.
@@strategictechnologist True. ChatGPT is good at that. I've learned in Gemini and Meta AI you DO need to copy and paste - which is sooooooo annoying! But the tradeoff to get up-to-date information is worth it to me with some things.
Dear Andy,
thank you for the video and all the work you do. It is super helpful- both technically and psychologicaly.
A little tip: the LLMs are still developed at a high pace. Therefore it would be helpful to define the version of the model you are using to do the test. For example say chatGPT 4o, or Claude 3.5.
All the best!
For the peer review rewrite task, I would retry Perplexity with the focus set to “academic”. 😊
I am SOOoooo LOVING this review!! I haven’t even finished watching it, but I paused to post this comment while ‘in the moment.’ I’ve been wondering about ChatGPT vs Perplexity for quite some time. I have paid versions of both and go back and forth between them depending on what I’m doing (I’m a health technology implementation scientist). Andy is being his usual hilarious self. I find myself chuckling through many of his videos, and I’ve watched A LOT of them. Andy, if you happen to see this comment, thank you so much for putting out such rigorous, provocative, instructive, and humorous videos. I can’t thank you enough for doing what you do.
Thank you soooo much for making the comparison. Spot on!
So…what model did you use in perplexity
chatgpt,perplexity said it to me 😂.
Great video as usual ❤❤. What video editor are you using? I'm noticing significant improvements in the editing.
Thanks Dr. Andy
In my experience, both hallucinate and cook up references that don’t exist despite the prompt being insistent on only papers that are real.
isn’t perplexity just google searching and summarising using chatgpt?
What did you use for recording??
#Objection
This is an unfair comparison: ChatGPT is an LLM, Perplexity in its advanced setting includes several LLMs including the ChatGPT4o itself (the same one). If you have set Perplexity with GTP4o you are comparing the same LLM, but if you set it to Academics, the results will give you papers in depth!
To access the advanced settings, you must clic your own profile picture!
You didn’t mention anything about the Perplexity “Pages” functionality or the “Playground” or “Collections”. You have more homework to do before a fair comparison can be made.
It may have slipped my mind, but as far as i understand, Perplexity is drawing on ChatGPT, Claude and more, and is not an LLM in it self ?
Perplexity is better at search
is the perplexity ai inside of ChatGPT same as perplexity by itself?so I don't have to pay twice :D
When you say Perplexity, do you mean Claude? Because Perplexity can run ChatGPT as well.
Did you test bang-for-buck? usage rates?
Thank you sir.
Hola! Would perplexity work in Spanish?
Yes it works in Spanish
Even before watching the video I knew you would say draw. You NEVER recommend one and stick to it. Perhaps that is how you make money. Stay well.
just goes to show you, prompt engineering is an actual JOB.
I have tried perplexity, not a fan. I prefer chatgpt with the custom scispace gpt
I'm already coming from your previous video, the one you recommended to watch at the end.. To be more accurate this one over here: ruclips.net/video/Ne-j_U2Vkfw/видео.html&ab_channel=AndyStapleton - So there you said that basically once I pay to Preplexity I can switch between LLMs from the settings. So right after watching this debate video it left me in many questions..
First and foremost, I really wonder which LMM did you pick for making this research debate.. ?
If ChatGPT 4.0 model was selected wouldn't it give the exact answers as the ChatGPT 4.0 Plus (web browser) would give.. ?
If you did set another LMM model for Perplexity (like Gemini or Llama 3 or else) and compared the results to what ChatGPT 4.0 Plus does, then you left me wondering why you did not mention it in the very beginning.. ?
I've appreciated your videos because I thought they are transparent and some what a true reflection of a real PhD student experiences but I guess it comes down to having more views and all that..
🕊
❤❤❤❤
Perolexity is getting worse and worse.
oh okay.
What made you say that?
The only problem is name perplexity and the founder CEO should act like a business man instead of nerd software engineer
I doubt many are going to care at all how the Perplexity CEO acts, providing he's delivering on his mission and providing a product that works.