Ep. #1013 - WTF is GPT?
In today’s episode of Startup Hustle, the Matts are back with a GPT discussion on deck! Matt DeCoursey and Matt Watson are here to dissect the ins and outs of ChatGPT and OpenAI. The hype is real—and we’re here to help you dig into the latest tech that got everyone talking.
Covered In This Episode
There’s a new thing that’s keeping everyone abuzz. ChatGPT and OpenAI are here to make life easier—or not. You get to choose the benefits it brings you based on its uses in your personal and entrepreneurial life.
In listening to Matt and Matt’s discussion about the basics of ChatGPT, you gain an understanding of what it is, how it works, and all the nitty-gritty facts.
Be updated with more than just a trend, but an extremely valuable tool. Tune in to Startup Hustle today!
- What is GPT? (02:23)
- How does GPT work? (03:18)
- Defining sentiment analysis (05:44)
- Limitations of ChatGPT (06:51)
- Writing songs with AI (09:45)
- Creating copy using ChatGPT (11:47)
- Challenges in creating new content (15:17)
- Using ChatGPT for strategies to avoid pitfalls as an entrepreneur (16:39)
- Asking ChatGPT about the top 10 startup failures (18:43)
- The founding team behind OpenAI (19:37)
- Other products outside ChatGPT from OpenAI (20:29)
- AI and machine learning—and what the future of AI looks like (22:20)
- On artists selling AI art (23:05)
- Using chat AI to create an onboarding document for salespeople (25:31)
- How AI chat is going to change the way people work (26:41)
- The big debate with AI (29:06)
- Use cases for AI in the real world (30:26)
- How to apologize to your podcast co-host (31:41)
- Writing code with AI (35:26)
- Using tech to increase efficiency between teams (37:00)
It takes a lot of time to come up with a list of 10 talking points to keep a 40-minute show. So we looked at this, and immediately our show producer, Jessica, thank you for spending so much time on set lists in the past. But congratulations on our discovery of GPT because the first thing she said when she logged two minutes was, oh my God, this is gonna make creating a setlist so much faster, easier.– Matt DeCoursey
What I love more than anything is the specific example you gave about the VC thing. It, somehow or another, intertwines all that into the story. It’s not like it scraped some shit off Google and threw it together. It’s smart.– Matt Watson
The key to all things in life these days isn’t about memorizing the answers. It’s being smart to know how to go find them. And this is just another way to go find them.– Matt Watson
Why are there 300,000 open software development jobs in the US? And the answer is we need to train and educate people with enough foresight to keep up with future demands. Which is very difficult to do as a society.– Matt DeCoursey
It’s hard to find expert developers—many entrepreneurs will agree with that. But that is what Full Scale is here for. Build your software development team quickly and affordably. It only takes two minutes to gain access to developers, testers, and leaders available for your project. Learn more on how to recruit skilled professionals easily!
Moreover, we also have podcast partners for additional business solutions you may need today. Be sure to take time to check them out!
Following is an auto-generated text transcript of this episode. Apologies for any errors!
Matt DeCoursey 00:00
And we’re back! Back for another episode of Startup Hustle. Matt DeCoursey here with Matt Watson. Hi, Matt.
Matt Watson 00:06
What’s up, man? Can’t wait to chat today.
Matt DeCoursey 00:09
This isn’t now. This is AI.
Matt Watson 00:12
Oh, God, in the matrix.
Matt DeCoursey 00:15
Well, it’s actually not. It’s me. It’s Matt. It’s Matt DeCoursey.
Matt Watson 00:20
That was, I know, you.
Matt DeCoursey 00:22
I know. Look, a whole world of listeners just have that same reaction. Yeah, so we’re taking a different approach with this episode. You see, it’s titled WTF is Chat GPT. If you have the internet, then you have probably seen people talking about Chat GPT. And what, like, their love for it. I actually made a comment that said I changed how I asked Facebook and how I changed my relationship status to that of being in a relationship with Chat GPT. We’re going to talk all about how I’m marrying AI and how I’m going to tell my wife about that gracefully. I’m going to ask if I have to tell her at some point. But yeah, we’re gonna do this whole episode. We do not have a setlist today. We will be creating it live through the use of open AI—Chat GPT. Before we do that, here is a quick reminder that today’s episode of Startup Hustle is brought to you by FullScale.io. Where we help you build a software team quickly and affordably. You go to FullScale.io to answer a couple of simple questions and find the matches that our platform gives you for software developers, testers, and leaders. So, Matt, Chat GPT made it to 1 million users in five days.
Matt Watson 01:39
Which was the fastest adoption of any technology ever built, right?
Matt DeCoursey 01:43
Crazy, and I’m positive it’s probably gone way past that because that was just five days. And then everyone’s talking about it now. I gotta tell you, the man in the world of AI, so Chat GPT, as I said, we’re creating this on the fly. So we’re going to start with, I asked him, I said, explain what Chat GPT is. And it actually says I’m not familiar with a model called Chat GPT. It said, “Are you perfect? Perhaps we’re referring to Chat GPT, which is the generative pre-trained transformer. It is a machine learning model developed by open AI that uses transformer arcade architecture to generate human-like text. It’s a type of language model that can predict the next word in the sentence based on the context of previous words.” It’s fine-tuned through various things in, man, I’ll get into that. I mean, it is a remarkably sophisticated bang, man. Like, yeah, what have you, what my initial impressions had been? Holy shit, wow. And also, man, this is gonna make some people obsolete.
Matt Watson 02:48
Well, it’s, you know, like everything. It’s a tool, right. And so I’m a software developer. And I’ve seen a lot of articles on LinkedIn and all this stuff about using it for computer programming. And one of the first things I did to it was I went to and said, Give me a simple C sharp code for how to read data from a database and update it. And it gave it to me. It spits it out. And the code was right, and I actually shared it with our team and sockets. And yeah, I shared it with the team at Full Scale, and I’ve never seen that in Slack. You don’t have to be like, Guys, this is crazy. And it’s not going to replace people’s jobs. But it’s a tool of like, you know, instead of like going and finding that somewhere else on the internet, it’s just like a productivity tool. But, like the computer programming stuff of it, I was super impressed, like what you could ask it to do, because it wasn’t just giving you random shit you would find on Stack Overflow or stuff. It was, like, actually, logically creating stuff like it was super cool.
Matt DeCoursey 03:48
So now, with that, I’ve tried tools like this before they’re out there. I’ve used them for years, everything from transcription software to just like aI content generators. And I gotta tell you, I have never, ever seen anything this accurate. And fast. I mean, it is like because most of the time I’ve tried these different things, and you’ll ask it to, like, write you some ad copy, and it’ll give you some paragraphs and stuff. And you have to go back and kind of delete two of them because they either sound like the other ones, or they’re weird, or you get to edit them. I have not had that issue. So, dude, back to creating this live, in the episode, I asked GPT what the best uses of GPT text generation language translation summarization and question answering sentiment analysis and text classification. Now some of those other ones do a great job of answering questions. I have asked a couple of factual historical things to give me a couple wonky answers. I asked about Full Scale software development, and it told me that I founded it in 2013 with a guy named Ian Belcher. I don’t know anybody’s name and culture and we did not find the company in 20. So that part of it was a lot all off. But yeah, but lobby questions answering stuff. Good chatbots. So what is sentiment analysis? I think that’s the hardest part of AI in general. It understands the tone. And it’s like it’s text like you can say things a lot of different ways.
Matt Watson 05:21
But the tone will indicate what if you’re like, well done now, well done. Well, that’s sentiment analysis.
Matt DeCoursey 05:25
I’m not telling you you did a job well, right. Well, sentiment analysis is super critical for all forms of social media, right?
Matt Watson 05:28
Like, that’s how they do a lot of moderation to figure out if it’s hate speech, or you know, all these other kinds of things that they’re looking for, to moderate social media. So a lot of social media companies have probably invested billions of dollars into various forms of that moderation.
Matt DeCoursey 05:50
It’s really critical AI technology, just as you pointed out, if you want to sign up for this thing, go to open ai.com. And you can just sign in. It’s free right now.
Matt Watson 05:58
It also does the really cool image generation stuff, too.
Matt DeCoursey 06:01
Yeah. Yeah, that is. I’m not as impressed with that at this point as I am with GPT. So I did just, you know, back to using this to create this live episode. I asked GPT, what are you not good at? And it says that the one limitation is it’s not able to understand the context in meetings of some words and ways that a human might. It relied on patterns in the data and was trained to generate text, not always producing output that’s completely coherent and accurate. So a lot of stuff does well. I have run into things where I’ll ask it a question. It does not access the internet to go out to find the answer. And then it also is pretty clear about the fact that anything that may have changed historically prior to 2022 it’s not going to be accurate about its build on old data. So well.
Matt Watson 06:51
Well, I think the dangerous thing here is if I go to Google, and I say like, you know, who founded Full Scale or whatever, your example, I get a bunch of results back and as a human, I go through them and I figure out what makes sense. Is this right? Does it answer my question or whatever, right? And then it’s up to me to kind of figure out whatever it is, and sometimes we were really good at serving up facts, and it seems very accurate. The problem is for us, if you ask this thing, that question, it doesn’t just give you a whole bunch of shit. And you’re supposed to figure it out, like serve it up. Like it’s fat, like it answered it, right. And that’s what’s dangerous about it.
Matt DeCoursey 07:23
So I did just Google who founded Full Scale, and it says Matt DeCoursey, at the very top of the page, and, like, 65-point font. I was Wow. And that came from CrunchBase. I would just be sorry. Sometimes you’re Matt. And sometimes you’re the other man. I’ve learned about that, right?
Matt Watson 07:40
Yeah, you’re right. It does come up as a real big deal like I’ve never seen. I’ve actually never seen Google give a response like that.
Matt DeCoursey 07:43
I’m like, wow, that’s like a 65-point font. Now, in that case, but actually, what I was expecting wasn’t that now I was expecting a whole page of search results, to have to possibly click a link to another page to go there. And you know, that’s one of the things about a search that is limited. Is that like you can, you’re gonna spend a lot of time kind of poking around for one good data source. And then sometimes it’s incomplete. And I don’t know, there’s just a lot to be said with that.
Matt Watson 08:12
I mean, Google has done it. Google has done a lot over the last few years to do those, like instant results like you ask something, and it gives you an instant answer. But you know, they must have humans on the other side that have picked one of those topics. They’ve kind of allowed it.
Matt DeCoursey 08:28
I don’t use humans for that. I think that’s an AI or machine learning-driven thing that probably ranks the quality of known data.
Matt Watson 08:37
Out of like we know.
Matt DeCoursey 08:40
It asks you for your feedback. It is just like a thumbs up or a thumbs down, do thumbs down and ask what the difference is.
Matt Watson 08:47
It just answers everything. Like it’s a fact, though. Right?
Matt DeCoursey 08:51
That was the versatility that shocked me. So you know, I’ve done we was Oh, man, if anything, this is just fun. So do we still have it? I asked you to write a song about startups. Yeah. And let me find that. So you know, what are some while I look for that. So first off, dude, writing a song is hard, isn’t it? Just a few seconds. So here we go. Will you want me to sing it? Rap? Yeah, yes. Okay, wrap it. In the beginning, just an idea in your head, a spark of passion, a dream to be fed. But now it’s growing, gaining momentum fast. It’s time to take that leap and make it last the chorus, which denotes startup startups. The future is bright with hard work and hustle. Anything is insight, innovation, and progress, the driving force behind startups, startups, a new world to find, and it goes on and gives the rest of the song, and it did it instantly. Like wow, like, that’s great. And then I was like, Man, I wonder how great this is. You actually did the same thing and had it written to the Twinkle Twinkle Little Star. So startups are shining bright in the world of business. You’re a site like, yes.
Matt Watson 10:03
My favorite was I asked her to do it in the style of Cardi B’s whap.
Matt DeCoursey 10:07
Yes, yes. So we were doing that. I was actually having a conversation with a well-known musician who I won’t name because I don’t want people to think his future songs are written by AI. But I was like telling him what I was like because writing songs is hard. I used to work in the music industry. I wrote a book about it. This is not easy. And this is making it look easy. So I was like, man, I was telling him I was like, I bet I could get this thing to write us a country song hit. And the first thing that came to my mind is like, Okay, why I asked it, I was like, write me a song about how I wish my beer was as cold as my ex-girlfriend’s heart. And dude, I’m telling you to introduce a head like I used to love you with all my heart. But now I see we were never meant to be. You were cold as ice, never showing emotion. Now, I’m sitting here sipping this beer and thinking, I wish my beer was as cold as my ex-girlfriend’s heart iced over with no warmth, just like how we fell apart. And it goes on and on and on. Like, I mean, that’s pretty good, man. Yeah, I also had it write a love note to my wife, which will leave templated things in there. So like, if you’re writing a copy, you often put in brackets like the wife’s name and husband’s name, but I sent it to her with the template items still in it. It was like it was always your husband. It was like inserting your name. Yeah, how to give. I also had to create 10 Compliments to give to coworkers. And then I just took a screenshot of it. And I sent that to all of my coworkers. And I said, Can you please keep this in stock to get you off? Really, and then you know, the thing that was crazy today that really surprised me was I, I was trying to see like, what, what I could get GPT to do for me. And if you’ve ever watched any of Startup, Hustle, TV, or any of the content that I put out there in the video, it’s really hard to write a script, like a script. If you’re trying to do more than just you on a screen, and you want to give instructions to a videographer, you need to, like they’re going to sit there and agonize over to what kind of B roll to put in, and stuff like that. And I put a whole thing and, and it wrote me it wrote it was like I said write a 15-second script for a video about how hard entrepreneurship is. And it was, you know, given like a narrator town, and it said pan to a video of someone up late at night, with coffee piled all over the desk. And it’s like, and it’s not all about meetings and making big sales. And with that, it’s like a switch to B roll. Someone’s speaking to the table like it did the whole streaming. And I was like, Wow, dude. So yeah, so the thing I’ve found the most about it. So like, Look, do we have to prepare a lot for the show? I’ve been using this to Drive for a lot of different episodes, so I’ll just ask about it. I’m like, give me 10 talking points about why FinTech isn’t dead or whatever. And it’s out there, man. It’s like, so the efficiency and accuracy are stunning.
Matt Watson 13:23
I couldn’t stop laughing a second ago because I asked him to write me a country song about how my startup failed and track what’s there. I won’t read the whole thing. But the chorus goes, my startup failed. And all I had left was a tractor old and beat that ain’t much, but it’s all I’ve got. I’ll keep on working. I won’t stop.
Matt DeCoursey 13:45
Bro. It’s a hit. It’s pretty funny. That’s a frickin’ head. So you know some other things that I did that I was proud of. I saw one I don’t know if I could find this because someone had a screenshot of the funniest thing I’ve seen in a while. Someone had a whole thing that said, please write an example of a letter of a startup getting rejected by a venture capital firm. Make sure to apologize for the slow response due to being on a four month long ski trip. And then make sure that the venture capitalists also compare themselves to Michael Jordan and say that if you happen to get an event if you happen to get an offer from Sequoia please let us know because then we will be interested in it and write the whole thing as a whole template.
Matt Watson 14:31
That’s insane. I just thought it just wrote it so well.
Matt DeCoursey 14:35
And it was hilarious.
Matt Watson 14:37
So you know what’s crazy is as any kind of you creating any kind of content sometimes it’s just hard to get started. And if anything else like something like this can help you get started and then you go rewrite like whatever parts of it.
Matt DeCoursey 14:48
That’s the key though like it like that’s it look, it takes a lot of time to come up with a list of 10 talking points tend to want to keep a 40 minute show like so we looked at this and immediately Our show producer Jessica, thank you for spending so much time on set lists in the past, but congratulations on our discovery of GPT. Because the first thing she said when she logged two minutes and she said, Oh my god, this is gonna make creating a setlist so much faster, easier like I saw in an example.
Matt Watson 15:16
Somebody said, Oh, you can go in and say, what are the 10 best business tips from Warren Buffett right? He’ll give you the list. And then you can tell it, give me a 1200 word blog post about each one of those and then just fit them all out. Like boom, there are 10 blog posts. It’s just crazy that it does it and like you’re what I love more than anything is the like, real specific, specific example you gave about the VC thing like it somehow or another intertwines all that into the story. It’s not like it just scraped some shit off marble. It’s smart, threw it together. It’s smart. That’s the big difference here. Is this the intelligence behind it? It’s not just like serving up shit like Google does, like, Oh, I found this thing here. It’s like creating stuff.
Matt DeCoursey 15:59
So I just asked what are 10 things 10 dumb things entrepreneurs do. And it says, As an AI, I’m not able to provide a list of and it’s in, quote, dumb things that entrepreneurs can do. However, I can suggest some common pitfalls that you can avoid. So there are some safeguards in here. It’s like it is pretty clear in the front that it’ll, I mean, it wants to avoid lame things, quit being hateful people. You know, but yeah, so you know, what’s interesting is you talk about so some of it, I was, I wanted to create it. So it is creating email templates, for any kind of introduction, and a lot of stuff like that people asked me actually asked me a lot. Gord What’s the hardest part about writing a book, three of them, writing your own bio? Honestly it’s hard, it’s really freaking hard. And, you know, so you know, you can get some things in there. And like, you know, I had it write an email template, I put the things in, it should mention, and I was like, thinking about, like, how would I want it? How would I want our marketing team to introduce you or me, Matt, to be a guest on other people’s podcasts? Right. And, you know, you get some people that aren’t familiar with writing templates like that, and they can sit around and you can be paying people to agonize over that for a while. And you know, like, I don’t know, do you by the way, are you curious about that? It won’t give me dumb things. But it tells you that the bad things not having a not having a clear business plan or goal, failing to conduct market research, not seeking advice or mentorship, not setting realistic goals, not managing your finances correctly, not paying attention to customer feedback, not being flexible and open to change, not protecting your intellectual property, not taking care of your team and not staying up to date with industry trends. I would say those are 10 pretty damn good examples of things you can do poorly to fail. Like? Yeah, by the way, there’s a whole bunch of detail after each one. I’m just gonna sit here and read.
Matt Watson 18:03
Well, I just asked it, what are the top 10 startup failures and why? And it listed out the 10 companies and gave, very succinctly, exactly why they failed. And it’s very, it’s fascinating. A couple examples. So the 10 companies like Theranos. Yeah, juicero, which was sold like a juicer, we work at Zenefits.
Matt DeCoursey 18:26
How’s we fail at work?
Matt Watson 18:29
Well, they, yeah, I mean, kinda, they didn’t fail. But I mean, you know, like, fab.com is on here. It’s like they were an e-commerce company that eventually had to lay off 1000 people and failed and shut down. Like segway is on here. There’s like, just a few different, like, different examples. There’s just, it’s just the ability to like, update itself.
Matt DeCoursey 18:49
So I can ask so many questions about FTX.
Matt Watson 18:53
Yeah, they will be on this list. For Shawn.
Matt DeCoursey 18:57
It’s just, by the way, do you know who the founder of this is?
Matt Watson 19:02
Matt DeCoursey 19:03
Matt Watson 19:05
No, it’s on.
Matt DeCoursey 19:06
He’s on the founding team. In Open AI, he is one of the founders there, man.
Matt Watson 19:12
I think he was one of the early investors.
Matt DeCoursey 19:15
Let’s ask it. Okay, who is the founder of open AI? Now, here we go. It is thinking and it is, it’s probably gonna waver because the founders are fighting over credit. If you’re right, open AI as a research institute and technology company focused on advancing AI and responsible and safe ways.
Matt Watson 19:46
Founded in 2015, by a group of entrepreneurs, including Elon Musk, and there’s a whole host of people on here, Sam, Walton, Elon Musk, and others.
Matt DeCoursey 19:49
So look outside of GPT open AI has a whole bunch of other stuff and its main body that has an API that you can use to connect to a whole world of other things. A whole Oral and other things. You talked about image generation, text completion. I mean, there’s a whole lot of stuff and you know, so I was looking at this and this is free right now you gotta assume it’s not always going to be free. They have some pricing plans for their other content and thought engines as well say, Dude, so affordable man. I mean, like you can produce. So here’s the problem. I did also make a post that said that if this was around, when I was in school, I wouldn’t have dropped out of five colleges. There’s some potential problematic stuff with this here because like, dude, I’m not kidding. I had it right. I said, Write me a 500 word essay about the moon. Nailed it.
Matt Watson 20:36
Yeah. Homework, you know, kids homework.
Matt DeCoursey 20:41
That’s how you use GPT. But honestly, the key to all things in life these days isn’t about memorizing the answers.
Matt Watson 20:46
It’s being smart to know how to go find them. Corral. And this is just another way to go find them, like, smarter, not harder, man. It’s no different than software developers being able to access StackOverflow. Okay, so now you can also use things like this that can help generate code for you and make you more productive. Love it. Awesome.
Matt DeCoursey 21:07
You know, Matt speaking to software developers, today’s episode is sponsored by FullScale.io, where you can go and build a team of experts, software developers, testers and leaders. It only takes a couple minutes to mean you do not need AI to fill out the very short list of questions that will help our platform match you up with the right people people ask me for like, Have you ever considered an AI or machine learning solution for your matching? It’s not really it’s actually not that hard. You ask people what they’re passionate about, you ask someone else what their need is and when to match up with them? Yeah, magic.
Matt Watson 21:40
Honestly, a lot of things that claim to be AI and machine learning are just basic algorithms.
Matt DeCoursey 21:46
Do you remember early 2019? And like no matter what you’re like, are you gonna get a salad with lunch? My machine learning algorithm says that I should get one and I should probably get creamy Caesar as a dressing. It’s like every dude, every single company we talked to there is a machine learning algorithm. I’m like, Dude, you do not have machine learning, nor do you need it. In an app that helps dogs at a dog park.
Matt Watson 22:12
Yeah, it’s great. Everything with AI. It’s crazy.
Matt DeCoursey 22:16
But dude, this is the future man. It’s here. It’s arrived. It’s arrived. The change log shit, man, it’s gonna change.
Matt Watson 22:25
I think the image thing is also interesting. I know. That’s not the main topic today.
Matt DeCoursey 22:30
But the daily diet, either daily or diet.
Matt Watson 22:34
Yes. There’s a few of those. And there are some artists that are up in arms about people selling like aI art on the marketplaces like the marketplaces are getting artless. I got a huge art collection.
Matt DeCoursey 22:45
I was in there playing with it and trying it out. I mean, dude, there’s something there. I mean, it’s kind of like NFT stuff, like the digital thing. There’s a big difference. Now, eventually, they’ll build a robot that could paint like my dude, Nate Trotter got to see the amazing paintings that he does, but I mean, so for example, though, it’s, it’s great for, like, if you’re writing a blog, like I want to put some images on my blog, you know, add images to it.
Matt Watson 23:05
And it’s like, I work in a company that does stuff at home services, like I can go on there and be like, I want a picture of a plumber fixing a toilet wearing a red shirt. And it’s like, Okay, I gotta, I got a picture that or it’s like, oh, brand colors, blue blue shirt. It’ll give me a blue. And, you know, it’s, it’s, you know, for things like that I could see it being pretty, pretty useful of like, sort of like stock photos and stuff. But it’ll do some wild shit. Like I went in there and I sent you a picture of Mickey Mouse wearing gold shoes, holding a banana playing chess on Mars. And it’s like, gives you a picture of it. It’s really crazy.
Matt DeCoursey 23:45
So now while you were talking, I asked GPT. I said, give me five reasons to ask my podcast co-host to stop talking. And it’s thoughtful, it’s thoughtful, it said, it’s important. So now it’s important that I communicate openly and honestly with you and to establish clear expectations and boundaries for this show. With that I’ve got a few potential reasons why you might consider not talking as much as you’re dominating the conversation and not allowing guests or me to speak and you’re speaking over or possibly interrupting frequently, you’re discussing topics or sharing opinions that aren’t relevant. We’re talking about GPT or matte not image creation. So right now and you’re not adhering to our agreed upon format or structure of the podcast which is I do all the talking and you ask AI how to tell me to stop talking Come on. Sorry. There’s one more reason you got to respect the time constraints, bro. Yeah, you’re causing the episode to run over on time.
Matt Watson 24:48
I asked for the best use cases for AI chat technology.
Matt DeCoursey 24:51
Oh, he’s pivoting the topic back. He doesn’t want to talk about I’m sorry. I just do it. This thing tells me just I Just but yeah, so did I really mean it? Okay, here’s some other things that I’ve done that Alright, so at Full Scale as we grow our sales team, it’s sometimes difficult for us to get people on board with the technical jargon and understanding related to the many, many, many, many, many things that we sell. And we do so I was able to create what came out to be about a 20 page document that is going to make new sales people immediately, like I asked it like, what does this front end developer do? What actually does a software developer do? What are the traits of a good one? What are the traits of a bad one? What should you look for? What is a front end developer was the back end of it was a mobile developer and then I literally had it I asked it to tell me what the 20 most popular code bases are, and then went through and individually asked it what all for more description on each of those dude and about an hour, I have a very comprehensive 20 page document that without this would have taken me days to create and have a good end result.
Matt Watson 26:01
It’s amazing. It’s incredible, man. It really is. It is really incredible.
Matt DeCoursey 26:05
It’s like wow, and then I got back to doing busy shit. More important shit like writing songs about startups and country heads. Love Notes to my wife 10 compliments for my wife. I did leave the template items in there. I got busted on that.
Matt Watson 26:20
Have you talked about AI chat? Have you been to McDonald’s recently? Where do you go through the drive thru? And you’re like talking to some AI robot?
Matt DeCoursey 26:33
No, I actually, it’s actually, yeah, they’re one buyer but more accurate than the people that were working there.
Matt Watson 26:36
Most of the time it is unless you get to some really complicated stuff. I feel like half the people that you go to take an order from that they spend most their time fumbling around, like how to use the computer to put it in, you know, and McDonald’s has a hard time finding people that want to do that job in a lot of places.
Matt DeCoursey 26:47
Yeah, I think it’s the key thing.
Matt Watson 26:54
Over time, I think AI chat. Like this will help with a lot of those things. It’s gonna be customer service.
Matt DeCoursey 27:01
That’s good. Well, as I mentioned at the beginning, I said some of this stuff is gonna put some people out of work. And you disagreed, I still believe it will, like it will, the Genesis like it’s like, it’s like calls. It’s like call center stuff. You know. It’s usually highly repeatable, somewhat low skilled jobs that most people don’t really want anyway, and businesses have a very difficult time staffing.
Matt Watson 27:25
I honestly would be very excited about this for customer service kinds of things. So for example, gig a book, you spent a lot of time helping, right? All sorts of help help documents about right forever. But users are lazy, they don’t want to go dig through all of those things. And so you can use different chat, and you put in, and it you put in the chat and it just uses some basic keyword matching to maybe find help documents that match. But now think about something like chat, GBT that could do with like, digest all of those help docs and all of a sudden be like an expert at all of it. And you can ask weird questions. And it’s not just like spitting out, like what it likes to do, you know word matching on it, like understands it and like, processes it with AI? And like, spits out? Like the exact answer, or suggestion, you know, to whatever the customer is asking like, that would be awesome. Because right now that keyword matching shit, that just gives you some random help docs Is not that helpful.
Matt DeCoursey 28:21
Right? Right. You know, but here’s the thing. And this is okay. So, you know, there’s this big debate with AI, you know, and they’re all Well, is it going to be like the Terminator? Is it going to take over? Maybe? I don’t know, man. But the one thing that I Okay, so when people are worried about that, is that commercially viable? Because really, in the end, all these things, anything, because capitalism, in my opinion, is the cure for so much. Why does NASA use SpaceX to launch rockets now? Because private industry is better at determining these needs and getting it done? And, you know, there’s, I believe, you know, if you don’t, if you don’t follow Matt Watson on LinkedIn, go do it. Because Matt makes some really compelling and interesting posts. But you had one the other day, you were talking about 10 million jobs open in the US, right? There are more jobs open than people that are available to look for the kids that are hireable. And you say why? Well, the thing is, if we’re not really it takes decades to why are there 300,000, open software development jobs in the US? And the answer is, we didn’t train and educate people with enough foresight to keep up with future demands, which is very difficult to do as a society. Well, I don’t want to get into the reasons why. Yeah. I’m not really talking about the politics or the sociological things of that. It’s just the reality of it. So you know, it really, in the end, businesses want to operate cleanly and efficiently and provide the best customer experience they cannot. And, you know, I was amazed because my cable went out the other day and I literally let my box break and I let it stay broken for like four days. Because I didn’t want to call I was like, God, I’m gonna fucking hate that.
Matt Watson 30:10
Matt DeCoursey 30:12
All AI it was all computer it was I didn’t speak to a human the whole time and it fixed it and like two minutes after I made it to where I needed to go and I was like, Oh, wow, okay, so that was a lot easier than I thought it was gonna be you talking about makes I have not done the McDonald’s experiment. But yeah, that’s, that’s the future of where this is going and like, but chatbots have been pretty shitty like until recently, they haven’t been that great.
Matt Watson 30:37
Like, they’re just like basic key words like they don’t really understand like, thought and logic and higher level thinking. And clearly chat AI or the open AI does like your example earlier about writing a letter to the VC with like, all these very specific details and complexities to the narrative. And it understands it like that’s, that’s the key is like, it’s actually intelligent. It’s just not an algorithm that’s matching on keywords.
Matt DeCoursey 31:01
The big thing I’m asking is to help me apologize to my co podcast co host for asking him to stop talking so much. Helped me apologize to my podcast co-host for telling. He was talking too much. Let’s see how long this takes.
Matt Watson 31:22
I wish I could get one that automatically chatted with my mom because she just wants to text me all the time.
Matt DeCoursey 31:27
Hey, are you available? Aren’t you Sure? Mom? What are you looking for? Oh my god, I can’t get my computer to work. Did you try restarting it?
Matt Watson 31:37
No mine. Either way your struggles are real. Yeah, I ran into your mom at the Apple store like a month ago.
Matt DeCoursey 31:39
And she was there seeking support? Oh, wait, we don’t have answers. We have answers. It’s important for me to communicate openly and honestly with you, Matt and apologize if I have cost in the offense. No, okay, good. But here’s some steps I can take to apologize to tell my co hosts that they were talking too much. I’d like to express my remorse for my words and actions now and acknowledge that I may have hurt your feelings. And I’m sorry for my behavior. There’s a whole list here. Now, here’s the thing: a man like this is some real shit in life that can help you out like, you asked it a real question. Do you remember? Do you remember years ago when I told you I even talked about it? I felt like it was like a billion dollar idea to have an app that figured out what your geolocation was. So wouldn’t do it while you’re sitting next to your wife or partner and send random compliments through tax yes. Do okay, because it came with if it came yet, but I mean, there’s some that could do that. And a lot of ways. By the way, I also want to take responsibility for my actions and not deflect blame or make excuses for what I said about asking you to not talk so much. It’s encouraging me to explain my perspective, ask for forgiveness and offer to make amends. So, but realistically, dude, like and it also gives me some advice. It’s important to be sincere and genuine with your apology and take steps to demonstrate that I’m committed to improving my behavior in the future. None of this is bad advice, man. It’s not like if people would ask GPT almost anything before they replied to in life that well, first off would make things a lot easier.
Matt Watson 33:38
So, I’ve got an answer for you.
Matt DeCoursey 33:40
Okay. You mean GPT house an answer? I’m not interested in your opinions. Not only the bot I only serve, I only serve the bot. I like big bots.
Matt Watson 33:52
I thought we were friends. I thought we had a bond. Now it seems like you don’t want to go on. You’re always yelling, you’re always pushing me down. I don’t know what I did. I just want to be found. My podcast hosts they’re being so mean. Had something. But it’s all been a dream. I tried to speak up but they just shut me down and started to think I should leave this town.
Matt DeCoursey 34:14
Oh my god. Oh, wow. That was really strong.
Matt Watson 34:21
Writing a song about my podcast host is being really mean to me.
Matt DeCoursey 34:25
Oh, that’s a song while the song is kind of a poem and plimmer guard but yeah. Well, I’m sorry that that happened to you, man. So yeah, man. I mean, overall, dude. Like, I don’t know. I gotta say this at least provided 35 minutes of entertainment so far during this recording. I mean, are there any, you know, as we kind of run out of time here? I mean, are there any other things that you’ve that were really impressive? I played with that, you know, I’m not a coder. I asked him a couple of simple things like, give me code to generate a random number. I was also talking to our product manager that runs the Full Scale platform. And I asked, I said, Give me code that will always produce the text. Julie is awesome. And then I sent the code. I sent a code to her. I’m like, can you insert this in as many places on the platform as you can find? And she actually sent me back a screenshot that showed it. So she put it in, like the dev environment and sent it back. And it was just kind of funny. But yeah, I mean, it was pretty useful in that regard. And yeah, I think you did some complex things. So you had some examples where like, SQL databases, and a lot of so it was, it was pretty simple stuff.
Matt Watson 35:30
I mean, so much about software development is, you know, integration of different things and implementing complex business logic and rules and stuff. And I don’t think it’s really going to do that. But you could have helped do some of the simpler pieces, like, Okay, how do I do this thing with the database right away, call this thing or whatever. I think that could be really useful. But again, it’s just a tool. It’s not going to replace the person doing the work, because there’s still a lot of, like, complex business rules.
Matt DeCoursey 35:58
And logic, but get a hub copilot attached to something similar?
Matt Watson 36:02
Yeah, it does some similar stuff.
Matt DeCoursey 36:05
And get some of those tools?
Matt Watson 36:08
I’m not sure if there’s some kind of similar thing that, for example, there’s something with GitHub that has gone through and taken like all the public code on GitHub, and it can do some suggestions of things.
Matt DeCoursey 36:20
It’s about trying to make things better, faster, cheaper, more accurate, like all of that. And, you know, okay, so let me ask you that, like, if you have an employee, would you rather have them use GPT? To create three outbound marketing templates? And have? Or are you like, No, you cheated? Or would you rather have them sit there for a way longer amount of time and probably produce less accurate or better results? Or do you care about the result?
Matt Watson 36:52
Because I’d rather go to chat with GBT and ask for five different versions, and then pick the best three and edit them a little bit and call it done.
Matt DeCoursey 37:00
Yeah. And by the way, that’s going to be a hell of a lot cheaper in terms of labor, focus, and others. So here’s the thing in your business, if you can create efficiency with tools like this, you know, we talked about the replacement of people, it doesn’t need to do that. You don’t have to be afraid of that it can let people be available to do more valuable things, like perhaps actually using that fucking template to reach out to people.
Matt Watson 37:24
Yeah, if anything, it should just increase productivity. But for some companies and people and maybe like, I don’t need to hire a copywriter or blog writer anymore, like maybe I don’t know. But the people who do blog writing full time, it should make them more productive, like they write faster and better output.
Matt DeCoursey 37:40
And like, honestly, here’s the thing is like there can be a level of accuracy at the Full Scale blog. Those are highly technical articles. And we spend a lot of time editing and doing quality control on that to make sure that the information that we’re putting out there is accurate, like, yeah, this will eliminate the need for that. That said, it’s not going to put it in WordPress, and it’s not going to create a great image with that. There’s a lot of stuff that can occur. But yeah, speaking of Full Scale, FullScale.io. Check it out. Software developers, testers, and leaders. You do not need AI to do that. Now, a couple things as we end our time here. I did not use AI to speak any of the words that I spoke today, but I’m hoping that comes soon.
Matt Watson 38:29
If I can replace you, that’d be great.
Matt DeCoursey 38:31
I know, working on AI. You know, I’ve said it a couple of times. I’ve asked people they’re like, I really love Startup Hustle. I’m like, cool. Who’s your favorite house? And sometimes they’re just saying that they’re talking to my face. So I’m like, are you talking about me? Are you talking about Matt, the other Matt? I don’t know. Congratulations on making it through 1000 episodes and five years of this shit with me.
Matt Watson 38:57
Yeah, thank you for making me do it.
Matt DeCoursey 39:03
If I had Chat GPT, that would have been so much easier for like four years, 11 months, and three weeks. So yes. Anyway, thanks to all of you, people are listening out there. Hope you enjoy this episode. And it was fun. This was fun, man. Maybe we just need to have a GPT chat.
Matt Watson 39:21
I do want to do it all the time now.
Matt DeCoursey 39:24
Do we have some performances to do for sure? I got a whole lot of musical instruments to do it. I’ll tune some stuff. What are you going to do? What are you going to play? Whether it’s going to be your ants? What drums are on? I just bought some of those for my son for Christmas. Perfect. Digital ones are quiet. Ooh, roll into my former employer. Do you know that Roland is the world’s largest maker of drums by sales volume? I’ve heard that. It’s crazy. It’s crazy because people think about making keyboards and stuff but no drums. Most people live in apartments and places. You can’t have a real drum. So I do want to point out that the very first thing we asked GPT was what Chat GPT is. It’s like, nah, ah, we’re just GPT generative pre-trained transformers. Would we have gotten that right, and 100 tries GPT stands for no? You didn’t need to try, bro.
Matt Watson 40:21
That’s why they call it GPT.
Matt DeCoursey 40:22
I just asked. Okay, if you have clever or fun uses for this, send us some examples. This was fun, man. I gotta get out of here because I’ve got 63 ridiculous question queues that have come up that I need to get into getting answered. So see you around, bro. Thank you.