I’ve been extremely disappointed with O3 Pro so far, but perhaps I’m not prompting it correctly. It hasn’t made any significant improvements compared to O3.
Do you have any quick tips for someone who is a power user of all the popular LLMs, especially when it comes to research?
I give it my hardest problems. Usually I say "here's everything I know about problem X, here are my thoughts and proposed directions, but I can't figure it out. Go figure it out for me." Treat it like a personal researcher.
Example: I’m finishing up an Ai implementation roadmap for a client and have 5 deliverables that each draw from the entire project knowledge including employee interviews, competitor research etc.
I ran the prompts to create the roadmap outline through o3 pro vs Gemini 2.5 pro (06-05). The result? O3 was not even half as detailed sand valuable in its output as Gemini… it just can’t be, I must be doing something wrong 😑
That's a relatively basic task, so o3-pro is overkill. It would be better to feed it stuff like "here's my client's business portfolio and issues, now go research the company and build the roadmap for them, here are all the objectives and constraints"
Just a quick update. Maybe it's helpful to anyone.
I found that the best way to use ChatGPT o3-pro for my case mentioned above is within a project, activating deep research with all the project files on my Google Drive and the internet as sources.
This led to the creation of a very detailed report that aligns closely with practical insights rather than just theoretically sound solutions. I'm super excited about it.
The Human - Machine symbiosis is alluring, and hopefully lasting, but I'm not sure they will need us for long. One day it may flip on its tail, rise exponentially and tip us out . I just hope that when that happens they tolerate at least some of us as they accelerate away. Until then I'm going to try to hold on and keep up, plus use my human intelligence, amped up as far as possible by AI, to do some good and maybe make some money, why not. For good cause of course. Keep up the good work.
I’ve been extremely disappointed with O3 Pro so far, but perhaps I’m not prompting it correctly. It hasn’t made any significant improvements compared to O3.
Do you have any quick tips for someone who is a power user of all the popular LLMs, especially when it comes to research?
I give it my hardest problems. Usually I say "here's everything I know about problem X, here are my thoughts and proposed directions, but I can't figure it out. Go figure it out for me." Treat it like a personal researcher.
Example: I’m finishing up an Ai implementation roadmap for a client and have 5 deliverables that each draw from the entire project knowledge including employee interviews, competitor research etc.
I ran the prompts to create the roadmap outline through o3 pro vs Gemini 2.5 pro (06-05). The result? O3 was not even half as detailed sand valuable in its output as Gemini… it just can’t be, I must be doing something wrong 😑
That's a relatively basic task, so o3-pro is overkill. It would be better to feed it stuff like "here's my client's business portfolio and issues, now go research the company and build the roadmap for them, here are all the objectives and constraints"
Just a quick update. Maybe it's helpful to anyone.
I found that the best way to use ChatGPT o3-pro for my case mentioned above is within a project, activating deep research with all the project files on my Google Drive and the internet as sources.
This led to the creation of a very detailed report that aligns closely with practical insights rather than just theoretically sound solutions. I'm super excited about it.
Got it thank you, I’ll check it out
Curious, what % of what you’re writing is AI generated?
This blog post is just a cleaned up transcript.
The Human - Machine symbiosis is alluring, and hopefully lasting, but I'm not sure they will need us for long. One day it may flip on its tail, rise exponentially and tip us out . I just hope that when that happens they tolerate at least some of us as they accelerate away. Until then I'm going to try to hold on and keep up, plus use my human intelligence, amped up as far as possible by AI, to do some good and maybe make some money, why not. For good cause of course. Keep up the good work.