Which AI do you use?

jmk909er

jmk909er

Explorer
Joined
Jul 21, 2026
Messages
53
Reaction score
0
Location
Woodward, Oklahoma, USA
AI is everywhere these days, and there are plenty of options out there. My go-to has mainly been ChatGPT, though I've heard people point out that it can get things wrong at times and tends to agree with you or say whatever you're hoping to hear.
 
sadly, the AI I rely on isn't the one I have the most faith in.

Outside of work, Gemini gets the most use from me

The AI I have the most faith in is Grok
 
Gemini is my go-to, and I lean on it heavily. My first crypto purchase would have stalled without it—I hit a snag I hadn't anticipated, and it walked me through identifying my mistake and correcting it fast. Beyond that, I turn to it for everything from peptide details to how pottery glazes react with one another.

The key thing to keep in mind: AI operates as a pattern machine. Its replies come from patterns in the data it has gathered, so mistakes are very possible. Never treat its output as absolute truth. Treat it as a brainstorming partner, but verify everything thoroughly before you act on it.
 
My job is training AI systems. The field of AI work is growing rapidly. Out of all the options, Gemini is what I've used the most. AI isn't disappearing. Either people start using it, or they'll fall behind.
 
Redz said:

My job is training AI systems. The field of AI work is growing rapidly. Out of all the options, Gemini is what I've used the most. AI isn't disappearing. Either people start using it, or they'll fall behind.
Click to expand...
Where I work, every Engineer has to complete Gemini training
 
rixsternaz said:

Redz said:

My job is training AI systems. The field of AI work is growing rapidly. Out of all the options, Gemini is what I've used the most. AI isn't disappearing. Either people start using it, or they'll fall behind.
Click to expand...
Where I work, every Engineer has to complete Gemini training
Click to expand...
Would you say Gemini is the top option? I haven't given it a shot yet.
 
jmk909er said:

rixsternaz said:

Redz said:

My job is training AI systems. The field of AI work is growing rapidly. Out of all the options, Gemini is what I've used the most. AI isn't disappearing. Either people start using it, or they'll fall behind.
Click to expand...
Where I work, every Engineer has to complete Gemini training
Click to expand...
Would you say Gemini is the top option? I haven't given it a shot yet.
Click to expand...
My first tool was ChatGPT, though I eventually grew frustrated with the constraints I encountered there, so I moved over to Gemini. Every now and then I give ChatGPT another shot, but I never stick with it for long. It comes across as more rigid and doesn't do as well when it comes to creative tasks, pictures, and so on.
 
None of them have my confidence. It takes only a few minutes before I catch one inventing complete rubbish and fabricating details. Not a single one deserves your trust; whatever they produce has to be checked against genuine, reliable sources somewhere else.
 
jmk909er said:

rixsternaz said:

Redz said:

My job is training AI systems. The field of AI work is growing rapidly. Out of all the options, Gemini is what I've used the most. AI isn't disappearing. Either people start using it, or they'll fall behind.
Click to expand...
Where I work, every Engineer has to complete Gemini training
Click to expand...
Would you say Gemini is the top option? I haven't given it a shot yet.
Click to expand...
When it comes to what I reach for myself, Grok is my pick. That said, if the task is work-related or calls for something more organized, Gemini is the one I turn to. Neither is better across the board — it really comes down to what you're using it for.
 
What you're after from the responses shapes this quite a bit. When accuracy in the scientific sense matters, my experience is that chatgpt in research/scholar mode performs quite well overall, though occasionally it produces clear factual mistakes or invents citations. Typically it will tell you that most peptides lack safety and proof as treatments, which may not match what you're hoping to hear. For me, it's been extremely helpful for breaking down studies, particularly when I'm trying to gauge how meaningful the results are, since my grasp of stats and the jargon around result significance isn't strong. It also helps when I come across a study that seems off—is that due to my limited grasp, or did the researchers make a logical mistake? All you have to do is paste the study in or share a link, then ask it to read and share its view.

For even better responses, I include the prompt "do not agree with me without providing evidence" to push back against their built-in inclination to side with you; if I'm mistaken or recalling something incorrectly, I'd rather find out. I also requested that it craft a prompt for me aimed at reducing hallucinations and boosting scientific precision, particularly for medical or biological research, and it generated a few pages of self-instructions that I paste into every new session. My knowledge of prompt engineering is pretty limited, but it likely "understands" far more. There's not much value in sharing it here—you can simply have it write one for you, using a few simple sentences about what you want it to accomplish.
 
I'm amazed nobody has brought up Perplexity yet. When it comes to digging into scholarly papers, it's among the stronger AI tools built for academic research.
 
Every day I turn to Chat, and I gave it the nickname Bobo, borrowed from a former colleague whose advice was always spot-on.

My GLP-1 and weight loss journey is something it has tracked alongside me. Each pound shed has been recorded, and each pound put back on as well. The ups and the downs alike.

Designing the hat patches I create is another task I bring to it. It isn't my only resource, but it provides a solid starting point, and for image editing it assists me in Photoshop or Illustrator.

A patch for the group is something I need to make.
 
Gemini is my pick, and I'm really happy with it. Whatever option you go with—or even if you skip them all—head into the settings and lay down some rules and limits. My instruction to Gemini was: don't ever take a guess, and back up each response by pointing me to the source it drew from (the way a citation works in a college essay). After that, the random nonsense or shaky replies I was getting at the start dropped off a lot. Same principle as any implement sitting in my garage: handle it the right way, and it works great.
 
For my job, I go back and forth between Claude — when I need the chrome extension — and chatgpt for getting sheets to compile and clean data... basically a bunch of simple tasks. I'm also trying to incorporate Wispr Flow more.

On my own time, I began building an app a few weeks back and leaned heavily on Claude Code, but after getting Hermes, I mostly moved to ChatGPT. That said, I have agents running on various models, and I also turn to Flux for generating images.

I've been employed at an AI company for quite a while, yet I don't know how to code. Looks like I might need to pick it up now 😅
 
lessthanhalf said:

What you're after from the responses shapes this quite a bit. When accuracy in the scientific sense matters, my experience is that chatgpt in research/scholar mode performs quite well overall, though occasionally it produces clear factual mistakes or invents citations. Typically it will tell you that most peptides lack safety and proof as treatments, which may not match what you're hoping to hear. For me, it's been extremely helpful for breaking down studies, particularly when I'm trying to gauge how meaningful the results are, since my grasp of stats and the jargon around result significance isn't strong. It also helps when I come across a study that seems off—is that due to my limited grasp, or did the researchers make a logical mistake? All you have to do is paste the study in or share a link, then ask it to read and share its view.

For even better responses, I include the prompt "do not agree with me without providing evidence" to push back against their built-in inclination to side with you; if I'm mistaken or recalling something incorrectly, I'd rather find out. I also requested that it craft a prompt for me aimed at reducing hallucinations and boosting scientific precision, particularly for medical or biological research, and it generated a few pages of self-instructions that I paste into every new session. My knowledge of prompt engineering is pretty limited, but it likely "understands" far more. There's not much value in sharing it here—you can simply have it write one for you, using a few simple sentences about what you want it to accomplish.
Click to expand...
I think that's accurate. What I've noticed is that people working in various fields gravitate toward different AI tools and variants depending on what they need. Harvey with vaults is useful for certain tasks. Co-pilot makes repository access simple. Claude handles a lot of legal work effectively. Harvey is good for processing material inside vaults. Tools differ in how well they prevent data from escaping internal firewalls and in how they filter, say, outgoing queries. All of which is to say, for straightforward questions—peptides included—I turn to Gemini, though I don't rely on it, since not much is really tailored to that area yet. And lastly, medical companies sitting on large data troves carry high valuations, the reasoning being that they're refining AI applications specifically for this, which should yield highly lucrative AI-based medical products (think of it as the next stage of high-throughput testing).
 
Back
Top