I know I’m late to the game…Chat GPT

[Anonymous]

Anonymous wrote:

Anonymous wrote:People don’t often realize that you can set the tone of your individual ChatGPT to be warmer or more severe. You can actually train it to be more clinical and not give you fluff, too. And it’s programmed to be sycophantic so you’ll use it more. Buyer (or free user in most cases) beware!

Even when not sycophantic, it is frequently confidently wrong. I have a deep expertise in one area and both Gemini and ChatGPT are very wrong some of the time when I ask questions in my area of expertise, but say their wrong answers with firm confidence.

It’s like the ultimate mansplainer. I have no idea why people trust it so much.

You’re probably using free version. If you work for a company that has a paid enterprise version there has been a big leap in accuracy and reductions in hallucinations. But you also can instruct it not to guess or hallucinate.

[Anonymous]

Anonymous wrote:

Anonymous wrote:I had a couple of abnormal results in a recent blood test, Chat GPT was actually pretty good for helping me understand the big picture.

The overall answer I got was "you should cut down on XYZ" but that my numbers are still actually pretty common for my age and at the moment have not much larger meaning than that.

Was that really helpful though? Of course we all know what we should cut down on.

Np. Chat gpt- the free version, I didn’t even use my co fancier version - flagged something in my meds causing a side effect that 3 doctors didn’t catch

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:I had a couple of abnormal results in a recent blood test, Chat GPT was actually pretty good for helping me understand the big picture.

The overall answer I got was "you should cut down on XYZ" but that my numbers are still actually pretty common for my age and at the moment have not much larger meaning than that.

I love ChatGPT for medical stuff. It's diagnosed 2 illnesses my kids had that doctors couldn't.

Now when we're sick, I send the symptoms to ChatGPT, get a diagnosis and suggested treatment, then hop on a Telehealth appointment, tell them what illness it is and what prescription I'd like, they write the prescription. SO much better than going to Urgent Care.

This doesn’t make sense to me. If nothing else, your doctors also have ChatGPT.

I doubt they’re using it fully yet. So many regulations and issues they need to deal before they can adopt it fully in appts.

I’m the one who went to 3 different doctors complaining of a side effect that was being exacerbated by my meds. No one caught it, and in fact referred me to other pricey time consuming testing. Honestly it makes me angry

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:People don’t often realize that you can set the tone of your individual ChatGPT to be warmer or more severe. You can actually train it to be more clinical and not give you fluff, too. And it’s programmed to be sycophantic so you’ll use it more. Buyer (or free user in most cases) beware!

Even when not sycophantic, it is frequently confidently wrong. I have a deep expertise in one area and both Gemini and ChatGPT are very wrong some of the time when I ask questions in my area of expertise, but say their wrong answers with firm confidence.

It’s like the ultimate mansplainer. I have no idea why people trust it so much.

You’re probably using free version. If you work for a company that has a paid enterprise version there has been a big leap in accuracy and reductions in hallucinations. But you also can instruct it not to guess or hallucinate.

No, paid. And I have instructed it not to hallucinate. I still catch very serious errors.

It is incomprehensible to me how much people blindly trust LLMs.

[Anonymous]

Anonymous wrote:What do people mean when they say they are asking chatGPT something? Do you just mean the Google search bar and there is an AI response? Or is this some separate website/program? Clearly I don't do this.

I occasionally read the AI response to my Google searches and wouldn't trust it at all. Somethings I know enough to know they are wrong on something. Sometimes I try to look for their sources and they either don't exist or they don't say what AI claims it says. Why are people using this thing?

There is a ChatGPT website and an app.
Think of AI like the internet a broad technology. ChatGPT is like one website built on that technology.
Ai is the platform ChatGPT is one application.
People use ChatGPT to draft emails, plan trips, get cooking ideas etc…
My last question for ChatGPT was how many drinks should I order for 100 person party.
It asked me how many adults vs children and for how many hours then gave me a shopping list.

[Anonymous]

Anonymous wrote:What do people mean when they say they are asking chatGPT something? Do you just mean the Google search bar and there is an AI response? Or is this some separate website/program? Clearly I don't do this.

I occasionally read the AI response to my Google searches and wouldn't trust it at all. Somethings I know enough to know they are wrong on something. Sometimes I try to look for their sources and they either don't exist or they don't say what AI claims it says. Why are people using this thing?

It’s a site owned by OpenAI. It helps with work (I use the enterprise version), planning menus and trips.

[Anonymous]

There are a million valuable ways to use ChatGPT. Those of you who are too dense to understand that are just demonstrating your ignorance when you advise people against using it or act like you know so much more than everybody else including ChatGPT. It's a tool and it is very informative and can be very helpful in analyzing both personal and professional or technical issues if you understand how it works. You have a lot of control over what type of tone it takes, you can personalize it.

[Anonymous]

Anonymous wrote:

Anonymous wrote:People don’t often realize that you can set the tone of your individual ChatGPT to be warmer or more severe. You can actually train it to be more clinical and not give you fluff, too. And it’s programmed to be sycophantic so you’ll use it more. Buyer (or free user in most cases) beware!

Even when not sycophantic, it is frequently confidently wrong. I have a deep expertise in one area and both Gemini and ChatGPT are very wrong some of the time when I ask questions in my area of expertise, but say their wrong answers with firm confidence.

It’s like the ultimate mansplainer. I have no idea why people trust it so much.

You are correct that AI is frequently wrong, and wrong on things that aren't hard to figure out.

I tell my friends and adult kids to choose a topic in which they are very familiar, almost an expert at. Ask a question about that topic and you will immediately see how many details AI gets wrong.

If you aren't an expert in that topic, you won't notice it's incorrect. AI makes it sound so assured on the slop it's spitting back to you that it's human nature to think it must be factual. But some of it is wrong. So now when google gives me the AI stuff first, I read it with a skeptical mind.

[Anonymous]

I’m surprised at the people inserting their private health information into ChatGPT.

The trust people put into tech bro creations surprises me.

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:People don’t often realize that you can set the tone of your individual ChatGPT to be warmer or more severe. You can actually train it to be more clinical and not give you fluff, too. And it’s programmed to be sycophantic so you’ll use it more. Buyer (or free user in most cases) beware!

Even when not sycophantic, it is frequently confidently wrong. I have a deep expertise in one area and both Gemini and ChatGPT are very wrong some of the time when I ask questions in my area of expertise, but say their wrong answers with firm confidence.

It’s like the ultimate mansplainer. I have no idea why people trust it so much.

You’re probably using free version. If you work for a company that has a paid enterprise version there has been a big leap in accuracy and reductions in hallucinations. But you also can instruct it not to guess or hallucinate.

No, paid. And I have instructed it not to hallucinate. I still catch very serious errors.

It is incomprehensible to me how much people blindly trust LLMs.

[Anonymous]

Anonymous wrote:

Anonymous wrote:I had a couple of abnormal results in a recent blood test, Chat GPT was actually pretty good for helping me understand the big picture.

The overall answer I got was "you should cut down on XYZ" but that my numbers are still actually pretty common for my age and at the moment have not much larger meaning than that.

I love ChatGPT for medical stuff. It's diagnosed 2 illnesses my kids had that doctors couldn't.

Now when we're sick, I send the symptoms to ChatGPT, get a diagnosis and suggested treatment, then hop on a Telehealth appointment, tell them what illness it is and what prescription I'd like, they write the prescription. SO much better than going to Urgent Care.

was it cheaper too - like no/minimum co-pay?

[Anonymous]

Just use Grok

[Anonymous]

As someone very knowledgeable about chatbots and LLMs, I can confidently say that using them is unethical and people who use them are losers who should feel shame for outsourcing their thinking to a corporate machine.

[Anonymous]

Anonymous wrote:As someone very knowledgeable about chatbots and LLMs, I can confidently say that using them is unethical and people who use them are losers who should feel shame for outsourcing their thinking to a corporate machine.

you also a vaccine denier? just curious.

[Anonymous]

Anonymous wrote:As someone very knowledgeable about chatbots and LLMs, I can confidently say that using them is unethical and people who use them are losers who should feel shame for outsourcing their thinking to a corporate machine.

Oh go away. As a programmer, I find them a vital tool. Does vibe coding work? Not really. But it is so nice not to have spend hours going down the rabbit hole to find the solution to a problem or trying to find any documentation about more obscure topics. And I no longer have to suffer the misogynistic twits of StackOverflow. Win! I am much more productive. Though as AI does keep taking leaps I do worry about the fate of the all the kids studying CS right now.