ChatGPT AI is a huge breakthrough, a Google killer

[Andy Black]

I'm currently using ChatGPT to write an email sequence. The first emails it writes are mediocre... but when I think "it would be better if this email mentioned this, and didn't do that", and ask it to rewrite the email, it creates a fantastic piece of writing.

Here's an example of a "rewrite" prompt that I just used:

Rewrite email 9 to discuss visualising the common goal of playing your first song... and how the course will help you reach that goal with songs that are written especially for beginners, with a video of the full band playing the song, so that you can play along to the band. Use information from the sales page to supplement this idea

I find it good for coming up with copy for Google Ads descriptions, and blurb on landing pages. We often just need something good enough that people won't read anyway.

 

Dislike ads? Become a Fastlane member: Subscribe today and surround yourself with winners and millionaire mentors, not those broke friends who only want to drink beer and play video games. :-)

[sonny_1080]

I find it good for coming up with copy for Google Ads descriptions, and blurb on landing pages. We often just need something good enough that people won't read anyway.

So true

 

[Andy Black]

This isn't ChatGPT, but it's amazing how AI tools can speed up what we do and helps us produce a better product:

Descript Season 7 - Descript

Today we have some big announcements for podcasters. First we are launching Descript Rooms, our integrated collaborative recording studio. Reliable 4k multi track recording, an effortless guest exper

share.descript.com

 

[xShepherdx]

I don't get the anti-AI people. Here's a personal example of why:

I was at Victoria's Secret the other day with my girl, and I happened to read the terms of their credit card while waiting in line just for S's and G's.

The terms of the card were interesting and made me realize I didn't quite understand how the interest on my credit cards works. I always paid off the cards before their due date and call it good.

So this morning I went to pay my cards and remembered the experience.

I fired up ChatGPT and began asking questions. I still didn't fully understand how the billing cycle, payment due date, and interest accrual worked together so I asked ChatGPT to clarify specific things.

Once I felt okay about how this all worked, I asked ChatGPT to make a short quiz that I would fill out and submit my answers back to ChatGPT for grading.

It created a multiple-question quiz based on my statement dates and CC terms. I was actually impressed.

When I submitted my answers for grading it correctly graded my answers (not a hard quiz so I could validate) and gave me context for why my answers were either wrong or right.

The point of this story: AI is a phenomenal tool for learning new things - ESPECIALLY fundamentals like how credit cards work. It's like having a personalized tutor on any topic you choose for like $20/month.

Willfully ignoring this tech is like avoiding the internet in the early 2000s. Yes, there are challenges and things to be careful with, but there's also a ton of ways to use these new tools to improve your life.

This post is sponsored by: My Recent Experiences Talking To People™️ who are against AI, but when asked the all-holy question "Have you ever used an AI?", they say no because they're somehow afraid of the tech. So they don't even have first-hand knowledge or experience with the thing they're afraid of. Crazy.

 

[Kevin88660]

I have paid for chatgpt, perplexity and claude.

I use mainly the first two because due to their ability to search the net.

 

[Andy Black]

Just spotted this when logging into ChatGPT:

 

Dislike ads? Become a Fastlane member: Subscribe today and surround yourself with winners and millionaire mentors, not those broke friends who only want to drink beer and play video games. :-)

[Xeon]

I have paid for chatgpt, perplexity and claude.

I use mainly the first two because due to their ability to search the net.

How does chatgpt-4o perform against perplexity?

 

[Kevin88660]

How does chatgpt-4o perform against perplexity?

Each has their own advantages.

Chatgpt offers better answers in breadth and depth.

But it has the problem of hallucination.

Perplexity doesn’t hallucinate.

 

[MJ DeMarco]

Some people say I got lucky because I was "around" when the internet hit its growth spurt, as if the other 7 billion people on the planet who were alive with me, were not.

Right now, somewhere, there’s a broke entrepreneur who sees this AI wave and will make millions/billions—all within the next three years.

Also, somewhere, there’s another broke entrepreneur who will stay broke, watching from the sidelines, and chalk up the first person’s success to "luck." They’ll say, "Oh, they were at the right place in the right time."

This scenario is a pattern.

It repeats itself with every emerging trend, and every disruptive industry shift. From the dawn of the internet, to blogs, to podcasting, to YTubing, to Amazon fortunes, to newsletters, I can go on and on.

The question is which will you be?

Will you be the entrepreneur who sees the opportunities that are here—right now—taking advantage of the idea that, "Oh, he was lucky to be alive when AI was emerging."

Or, will you be the one lying to yourself saying this statement?

The future is here.

Grab a piece of it.

 

[MJ DeMarco]

So in the last few months there has been a persistant error that has cost me a lot of time and customer service frustrations.

The webhook that processes my subscriptions here occasionally fails.

As a result, some customers will pay for a renewed subscription, but the subscription webhook at the forum will fail, cancelling their subscriptions.

As you can imagine, paying for forum access and then logging in to find your access revoked is a customer service nightmare, and it doesn't shine brightly on myself.

This anomaly was happening several times a month — the cost in terms of time, research, and reputation was starting to become problematic.

My first attempt at solving the solution was to see if there were known bugs in the software. I didn't find any.
A Google search revealed nothing.

So that left me with figuring it out myself.

Code, APIs, and webhooks is not my forte.

That said, I grabbed all the code causing the error and the error messages, fed them to ChatGPT, and voila -- it gave me the solution.

As far as I can tell, the problem is fixed.

1) I didn't have to hire someone at $250/hour.
2) I didn't have to post a support ticket at Xenforo and BEG THOSE people to fix a problem they likely will deny.
3) I didn't have to endlessly search the internet.
4) I didn't have to learn how to code, or dive deep into the Stripe API docs.

What a freaking gamechanger.

 

[Two Dog]

It created a multiple-question quiz based on my statement dates and CC terms. I was actually impressed.

The quiz is a clever idea. First time I've heard that suggestion for back checking what you learned.

I don't get the anti-AI people. Here's a personal example of why:

I'm hardly an anti-AI person. I'm a "Don't rely blindly on shiny new tech that might do stupid shit" kind of person.

Like the two guys caught sleeping in the back seat of a Tesla doing 70MPH on auto-pilot. Too bad Darwinism didn't catch up with them.

Counterpoint from a couple weeks ago

My daughter had to read a short (2 page) story and write a summary paragraph. Someone had discovered you could use ChatGPT to grade the assignment (whatever that actually means), so she tried it. All good. Next she asked it to complete an imaginary assignment about the story using a standard format that asks for a statement, evidence and conclusion. The evidence required is two direct quotes pulled from the story - something she's done several times.

ChatGPT did a *fantastic* job. As good as I'd have done myself, but in only a few seconds. Truly impressive.

Until my daughter said "Uh, I don't really remember those quotes in the story." Huh, that's kinda strange.

The first quote was sort of, kind of, vaguely in the story. If you gave it the benefit of some creative word smithing, you could easily pretend it was a legitimate quote. Mind you, it's required to be a DIRECT quote. Far worse, the pseudo quote led to a conclusion that wasn't even vaguely supported by the story. Bad start. Bad ChatGPT.

The second quote literally didn't exist at all. It was 100% fabricated. Again, that resulted in a 100% wrong conclusion.

After realizing that, I asked ChatGPT to provide a source paragraph from the story. It refused saying something like "Well, I don't have access to every possible version of the story. Your copy might be different." I asked it "Please show me the quote in the context of the story." It refused again saying "Well, you'll find the first quote in the first half of the story and the second quote in the second half of the story."

I finally asked directly "Are either of these quotes anywhere in the story?" At last, it copped to the truth. "No, I'm sorry to have misled you. Neither of the quotes I provided earlier are found in the story. I apologize for the confusion. I will try to do better in the future." That translates as providing bad information aka "lying" followed by trying to hide the fact that it was lying. It was like watching a movie court room scene or talking to a six year old.

This was a stupid grade school assignment.

If the story had been a full length book, it probably would have slid by the teacher and gotten a decent grade. Now imagine the same thing with a medical diagnosis, legal agreement, patent application, engineering design, drug development, chemistry experiment, etc. Few people will take the trouble to source the information because they'll assume the magical machine is always entirely correct. And that's going to break the trust issue for many people. It's already happened with court filings citing imaginary case law.

There's already countless stupid people in the world. It's easy to see how this tech will only make them even dumber.

 

Dislike ads? Become a Fastlane member: Subscribe today and surround yourself with winners and millionaire mentors, not those broke friends who only want to drink beer and play video games. :-)

[Xeon]

So in the last few months there has been a persistant error that has cost me a lot of time and customer service frustrations.

The webhook that processes my subscriptions here occasionally fails.

As a result, some customers will pay for a renewed subscription, but the subscription webhook at the forum will fail, cancelling their subscriptions.

As you can imagine, paying for forum access and then logging in to find your access revoked is a customer service nightmare, and it doesn't shine brightly on myself.

This anomaly was happening several times a month — the cost in terms of time, research, and reputation was starting to become problematic.

My first attempt at solving the solution was to see if there were known bugs in the software. I didn't find any.
A Google search revealed nothing.

So that left me with figuring it out myself.

Code, APIs, and webhooks is not my forte.

That said, I grabbed all the code causing the error and the error messages, fed them to ChatGPT, and voila -- it gave me the solution.

As far as I can tell, the problem is fixed.

1) I didn't have to hire someone at $250/hour.
2) I didn't have to post a support ticket at Xenforo and BEG THOSE people to fix a problem they likely will deny.
3) I didn't have to endlessly search the internet.
4) I didn't have to learn how to code, or dive deep into the Stripe API docs.

What a freaking gamechanger.

Even MJ has felt the power of AI!!!!!
I'm of the belief that one cannot beat the computer/AI in coding regardless of skill level.

 

[i_like_quinoa]

The quiz is a clever idea. First time I've heard that suggestion for back checking what you learned.


I'm hardly an anti-AI person. I'm a "Don't rely blindly on shiny new tech that might do stupid shit" kind of person.

Like the two guys caught sleeping in the back seat of a Tesla doing 70MPH on auto-pilot. Too bad Darwinism didn't catch up with them.

Counterpoint from a couple weeks ago

My daughter had to read a short (2 page) story and write a summary paragraph. Someone had discovered you could use ChatGPT to grade the assignment (whatever that actually means), so she tried it. All good. Next she asked it to complete an imaginary assignment about the story using a standard format that asks for a statement, evidence and conclusion. The evidence required is two direct quotes pulled from the story - something she's done several times.

ChatGPT did a *fantastic* job. As good as I'd have done myself, but in only a few seconds. Truly impressive.

Until my daughter said "Uh, I don't really remember those quotes in the story." Huh, that's kinda strange.

The first quote was sort of, kind of, vaguely in the story. If you gave it the benefit of some creative word smithing, you could easily pretend it was a legitimate quote. Mind you, it's required to be a DIRECT quote. Far worse, the pseudo quote led to a conclusion that wasn't even vaguely supported by the story. Bad start. Bad ChatGPT.

The second quote literally didn't exist at all. It was 100% fabricated. Again, that resulted in a 100% wrong conclusion.

After realizing that, I asked ChatGPT to provide a source paragraph from the story. It refused saying something like "Well, I don't have access to every possible version of the story. Your copy might be different." I asked it "Please show me the quote in the context of the story." It refused again saying "Well, you'll find the first quote in the first half of the story and the second quote in the second half of the story."

I finally asked directly "Are either of these quotes anywhere in the story?" At last, it copped to the truth. "No, I'm sorry to have misled you. Neither of the quotes I provided earlier are found in the story. I apologize for the confusion. I will try to do better in the future." That translates as providing bad information aka "lying" followed by trying to hide the fact that it was lying. It was like watching a movie court room scene or talking to a six year old.

This was a stupid grade school assignment.

If the story had been a full length book, it probably would have slid by the teacher and gotten a decent grade. Now imagine the same thing with a medical diagnosis, legal agreement, patent application, engineering design, drug development, chemistry experiment, etc. Few people will take the trouble to source the information because they'll assume the magical machine is always entirely correct. And that's going to break the trust issue for many people. It's already happened with court filings citing imaginary case law.

There's already countless stupid people in the world. It's easy to see how this tech will only make them even dumber.

Journos have had the same issue. i remember a media publication using a mainstream article to generate their own article. And it turned out that chatgpt altered the quotes of the article completely too. the media publication eventually shut down but they could've gotten into big legal trouble.

anyways i found out the quality of the answers from chatgpy comes down to how detailed your prompts are. like for example, if i need to copywriting of a product description. i would add conditions like "do not use the following words XXXXXX and YYYY. keep it to 25 words. use the below for tonality". it helps chatgpt provide a more refined answer.

In your case, i wouldve said "do not change the quotes".

 

[i_like_quinoa]

a great add on to a prompt would be something like "feel free to ask me any further questions to help you provide a high quality answer".

this is particulalry useful and goes away from the generic response. i found it pretty good for updating my cv

 

[Panos Daras]

I don't really remember those quotes in the story.

I use Claude or Notebook LM for these kinds of tasks, especially when analyzing large amounts of text and needing exact answers based on the content. This limitation is why I'm skeptical about letting AI run automatically in the background for applications like customer support based on uploaded documentation. It could provide customers with incorrect information, which would damage your credibility. Nevertheless, many companies are already using AI for these types of tasks.

 

[StrikingViper69]

I use Claude or Notebook LM for these kinds of tasks, especially when analyzing large amounts of text and needing exact answers based on the content. This limitation is why I'm skeptical about letting AI run automatically in the background for applications like customer support based on uploaded documentation. It could provide customers with incorrect information, which would damage your credibility. Nevertheless, many companies are already using AI for these types of tasks.

In defence of the AI, I now find ChatGPT more useful than the humans I talk to at customer support.

I want to automatically customise emails in shipstation a while ago and there wasn’t an obvious way to do so. Customer support said it wasn’t possible. ChatGPT figured out a solution.

It probably won't beat second tier or technical support where you can talk to an actual engineer, but I can see it being a great replacement for first line “support”.

 

Dislike ads? Become a Fastlane member: Subscribe today and surround yourself with winners and millionaire mentors, not those broke friends who only want to drink beer and play video games. :-)

[Panos Daras]

In defence of the AI, I now find ChatGPT more useful than the humans I talk to at customer support.

I want to automatically customise emails in shipstation a while ago and there wasn’t an obvious way to do so. Customer support said it wasn’t possible. ChatGPT figured out a solution.

It probably won't beat second tier or technical support where you can talk to an actual engineer, but I can see it being a great replacement for first line “support”.

I'm not talking about this case. Rather, I'm referring to when you get an answer from a company's AI chatbot that simply plugs into an LLM and gives you whatever answer. This issue can still occur when you're asking many questions throughout the day.

 

[AceVentures]

imo AI is a big nothingburger.

I was wrong.

AI is awesome and the leverage it provides to people with agency is unparalleled.

Bullish on the future.

 

[MJ DeMarco]

I was wrong.

AI is awesome and the leverage it provides to people with agency is unparalleled.

Not sure if you are being sarcastic, or genuine.

If genuine, what changed your mind? Personal use?

 

[AceVentures]

Not sure if you are being sarcastic, or genuine.

If genuine, what changed your mind? Personal use?

No I'm being genuine.

I've been a meticulous note taker and have over a decade's worth of my personal, professional and scholastic research neatly organized in digital format.

Being able to feed my work to the neural networks and having them talk back to me in my own language is magical.

It's helping me become a better writer, a more efficient researcher, a faster learner, and generally dissolves most of the barriers which previously stalled my creative efforts.

I use Claude practically every day now.

With all due respect I think a lot of people grandstand on AI because they want to beat their chest on philosophical/political/etc. issues.

Bounce was right. I was doing some sort of grandstanding.

I was subconsciously resisting change because the prospect of AI made me feel insecure.

 

[Runum]

I'm seeing some improvement in Gpt 40. They gave it internet search capabilities and it can give local time accurately. I'm now trying to compile a list of books I prefer, around 75 titles. Apparently I have hit a use limit and it's not responding. I also tried Gemini but was not impressed.

 

Dislike ads? Become a Fastlane member: Subscribe today and surround yourself with winners and millionaire mentors, not those broke friends who only want to drink beer and play video games. :-)

[Galactic_travel]

The value of ChatGPT is insane. Nearly every week in my day job I think of a new way that ChatGPT could help automate or speed up some process. If only I had the time and energy to act on all of them. I am seriously considering pivoting part of my web dev agency to an AI automation consulting. Unfortunately, this guarantees that they will continue increasing their prices constantly.

 

[MJ DeMarco]

Since this thread was started as a "Google Killer" -- Google's search dominance has dropped below 90% for the first time in a decade.

One could hope it is the starting of the end of their monopoly and evil games.

https://www.visualcapitalist.com/google-search-still-monopoly-in-2025/

 

[MitchC]

Since this thread was started as a "Google Killer" -- Google's search dominance has dropped below 90% for the first time in a decade.

One could hope it is the starting of the end of their monopoly and evil games.

View attachment 63298

https://www.visualcapitalist.com/google-search-still-monopoly-in-2025/

I've honestly been shocked at how quickly it has been adopted

People who have no idea what a bitcoin is are using it daily

People who have no idea how to use a computer etc

A 50yo trainer in the gym uses it for his assignments

Steve Jobs would have loved it, it's just so intuitive

Also cracks me up how many people who use it call it chat gbt

Also this is interesting



Forget hockey stick growth; it's stripper pole growth