A podcast by 37signals about the better way to work and run your business. Hosted by Kimberly Rhodes, the Rework podcast features the co-founders of 37signals (the makers of Basecamp and Hey), Jason Fried and David Heinemeier Hansson sharing their unique perspective on business and entrepreneurship.
Welcome to Rework, a podcast by 37signals about the better way to work and run your business. I'm your host, Kimberly Rhodes. Okay. I have to admit that I know very little about AI or artificial intelligence. I have not used Chat GPT, but I hear about it constantly. And if you're a small business owner, you may be wondering when you need to be concerned about it. Do you need to jump in immediately? Is it going to take over your business? And if you're like me, are you already behind? Here to talk about it are the co-founders of 37signals, Jason Fried and David Heinermeier Hansson. I'm excited to hear your thoughts on this as entrepreneurs in the tech space. I'm sure you guys have some thoughts. First, have you guys jumped on board with this Chat GPT situation?
Well, first, Kimberly, you're not falling behind. This is the one thing I want to, I wanna say that really kind of bugs me is, is this sense that like people feel like they're falling behind. First of all, this brand new, um, it's in flux, it's fascinating, it's interesting. Um, but no one's falling behind. And if you're, if you're on LinkedIn, everyone's, you're falling behind cuz your marketing is falling, be everything's falling behind. And I, I don't like this fear-based, this is an amazing new interesting thing. Like, and I don't really, we don't really know. I, we, no one really knows where it's gonna end up going. So I would meet it with curiosity and not with worry and fear. And I, I don't like the, the, the themes that are out there right now. So don't worry about that. It's interesting. It's, it's very interesting.
It's very impressive. It's, and the cool thing, by the way about it is that you, you don't have to really know what you're doing. You can just talk in a sense or, you know, write simple human questions. And of course there's techniques on how to get more out of it and whatnot and there's other things it can do, but for the most part, if you just meet it with curiosity and you ask some questions and you're just plain language talk with it, it's like there's not much to learn to begin with. So that's the first thing I wanna say. Cause I know you led with the fear thing and everyone else is, and it and, and you're, I feel like it's, it's infecting people. They feel like they're, there's FOMO already and, and there shouldn't be. So I just wanna start with that.
The FOMO Fest is out of control . Yeah, I am actually, it is to the point where I have to arrest myself, not slandering all of AI with this incessant annoyance I have with "here's five revolutionary AI tools that just popped up this week. If you miss him, you're dead!"
Dramatic. Um, and there's just something that's almost a parody. Like if you were gonna make a parody of the tech industry and how it just gets, it's itself wrapped up into this hype cycle. You just have to reprint actual tweets and actual LinkedIn posts from like the last week. And it's really annoying because there's actually something there, there. I mean ask Jason, I've played a bunch with both Mid Journey for images and with Chat GTP for text and it is probably the first time in a long time where I've seen something brand new that's already, wow, in a sense where you can not just see that the future's gonna look different. I've had that experience with other things. I've had that experience with virtual reality. For example, the first time I tried VR I went like, wow, in 10 years that's gonna be incredible.
This AI stuff is incredible now and the rate of change is really high. But that also means that like there's nothing to catch up because you know what, that rate of change currently at least is not slowing. So anyone who learned everything there is to know about AI, at least according to all these tweets, they'll be out of date next week anyway. So if you were just laying on the couch, you were sitting it out, you were being a late adopter, you can jump right into the pool next week and catch up right away. The sort of half-life of knowing everything about AI currently is like 48 hours. So that's actually a good thing if you're not, um, sort of following everything. If you'd follow all the Mid Journey updates, like we're on mid Journey 5.1, if you were an expert in Mid Joney three say you knew just the right trick to make sure that the hands coming outta mid journey actually had five fingers and not seven and an alien eye on it.
You know what? That's all for. Waist mid journey improved itself. Now it can do hands and they're great. So it is a FOMO fest. It's really annoying. It should not let you be discouraged about actually trying it out. As Jason says, this is in some way I think actually the hype is proportionate to the fact that there is no skill. There is virtually no skill currently involved with using AI well. That is the magic. It is not a programming language. The whole magic of this is that Jason says you're just asking it questions in plain English and whatever tweaks you can come up with or fine tuning the last percentage, it doesn't matter that last percentage has vaporized next week as we talk about. Right? So I think this is why everyone is freaking out so much because the floor is so low. I could hand chat GTP to my oldest son and he'd know how to use it even if he hadn't seen it at all.
That also means we don't know where this is going. We don't know what it's gonna do to the economy. We don't know what it's gonna do to jobs. We don't know what it's gonna do to a bunch of things. There's some niches at the, the tail that are being impacted right away. I'd probably be perhaps a little worried if I was a stock photographer, right? Saying, well, like that's perhaps one of the ones that's earliest in the line for API overlords to take over. But I saw a great tweet yesterday which essentially said, do you know what, if your job wasn't outsourced, it's probably not gonna get stolen by AI tomorrow. Now that's not an encompassing, uh, statement that qualifies everything. But for example, in programming where I've actually used chat GTP a fair bit, I think the hype is totally outta control. People are acting as though you can just tell, oh, make a great web app and it'll produce that.
Yeah, it'll produce something and then if you need to change it and you don't understand what it's actually produced, do you know what you're barely better off. Again, this may improve and these tools might get so good that they can rewrite the code that they have and so forth. But right now they feel a little bit like code generators. Code generators can save you a lot of time by dealing with a bunch of boilerplate that you otherwise have to look up their arcane syntax off and they can have speed that up. But if you don't understand what's being produced, you'll hit the wall relatively quickly in terms of what you can do with it. So this is also where we talk about AI as though it's this monolithic thing, which I don't think it is at all. Um, we talk about AI in legal as though it's the same as stock photography, which it's not at all.
There was just this hilarious case of someone using chat GTP the other week as part of their legal brief and Chat GTP hallucinated, cited a bunch of cases that never happened and the judge of course was um, fairly miffed with the fact that we just had outright lies in official depositions versus if Mid Journey makes up, uh, whatever fantastical image that looks totally real, but isn't no one at least in most cases are getting harmed I guess unless it's the Pope wearing a fancy, um, designer jacket and people are freaking out over that. So that has its own. But I just think like embrace that. How often do you get to live during a time of this much uncertainty? This is sort of like what happened with the internet, but the internet took a good at least five years to ramp up. We've gone like internet levels of, I hate the freaking word but I'll say it anyway, disruption in like five months. That's just fun. I mean, what a time to be alive.
I want to pick that word out cuz I think this is the key. Just have fun with this right now. And it is fun. It's actually fun. Um, and this is what pisses me off the most about how companies are already implementing this. They're taking it so seriously. It's like, well you can summarize this long presentation that was already boring as shit. And you get a summary of it and, and it's like so serious. It's like the whole thing is this is fun. Mid Journey is a blast to play with, Chat GPT is a blast to play with. Like, I'll write something sometimes I'll throw it in there and go give me a rhyming headline that would be fun for this. Like, you know, it's like you, you kind of wanna smile at the responses, you're impressed by them, but you also wanna play with it.
And that that's why like everyone's like, well this will, I mean, basically the big, the parlor trick right now is like summarize like all the products that have implemented, uh, I shouldn't say all, but most of the business tools that have been implementing AI, it's like, write this for me are, summarize this for me. It's kinda like the most boring possible implementation of that. And is it useful? I don't know. If the work wasn't necessary in the first place, the fact that you can do it faster. So what? That's the thing that's so frustrating to me. Um, I think we are on this precipice of this incredible tech that's also really fun to mess with and really fun to bring your curiosity to. And it's sort of being immediately co-opted by the most boring aspects of business, communication specifically. Um, so anyway, that that's, that's something that bugs me about it.
But um, as far as like people are always wonder, are you gonna bring it to your stuff? Like at some point I'm sure we'll bring something in when it feels like it makes sense in a way that's not just the same thing everyone else is doing, which is essentially summarizing and, and pre-writing things. Um, we have begun to experiment in HEY with it being able to write a response for you. This is a staff only thing on our side, we're seeing if it's even useful or not. So there's some of that that we're playing with a little bit, but we're not rushing to add this to our products right now. I think there's plenty of time, we're very early, things are gonna change. Who knows what novel, what novelties are gonna pop up in the next year. And when it's time when it seems like it makes sense, we will, we will bring it to life. But until then I think we're just gonna kind of stand by and watch and enjoy.
I think part of what's fascinating here is we don't even know the final form factor. Like does it even make sense to put it into products or is AI just gonna be like you have your own AI and that's an app that runs on it on its own. Is it gonna be more like Her? It's actually a spoken interface that you deal with entirely. I think Jason, you're post contrasting Apple's new, uh, AR goggles. The Vision Pro versus this AI model that is mostly natural language is really interesting is that for the first time you can see the outlines beyond the smartphone. The smartphone has been the dominant computing platform for a good, whatever, 15 years now at least. And suddenly now it's up in the air and you could imagine that the next interface is actually spoken because it'll actually be good. Or you could imagine that it's gonna be something like those AR goggles.
What's interesting to me is, as Jason says, the business industry doesn't even know what it wants AI to be or do. It's taking the most boring way of dealing with it, compressing business speak in and out to humans, right? Humans wanna write something short, they wanna read something short, but we've somehow convinced each other that you have to write something long because that's the proper thing to do and then someone can be bored on the other end trying to pick out the point you're actually making. And now we're injecting AI to remove that. I think that's just a hilarious critique of business writing in general, but I also think it's emblematic of what Chat GPT actually produces. If you look at some of these like, hey, write me a letter asking someone for something, right? It writes that awful business language verbose bullshit. It uses three paragraphs when two lines would do.
Now is it impressive that it can, and that it actually sounds like a corporate drone perfectly to the pitch? Yeah, really impressive. Is that what I want to receive? Absolutely not. And I think we've actually already seen that in some areas of the internet that, um, Stack Overflow for example, I saw just banned chat p answers because they usually have this character of overly verbose bullshittyness to them. Again, doesn't mean there isn't some magic here. And part of the magic is the fascination that like, oh, corporate business speak is something you could automate. I mean, again, isn't that a hilarious critique of corporate business speak? So let's just see where the, where the chips fall, give it a give it a year. In fact, to some extent, as exciting as this is, I wish I could hit the fast forward button and jump over this whole FOMO fest and just arrive at like, okay, here's some actual useful shit you can do with it.
And the form factor's relatively known and we don't all have to be these thin layers on top of open ai. That's the thing that gets me about all these AI startups and all these AI features they're presenting as, as they're doing some, they're not doing anything. They're calling an API. Like it's the most bare bones implementation is almost all the cases, which means that they are picking up pennies in front of the steam roller. And that has happened already to about a hundred AR startups, if not more. Open AI would, I actually did this, I bought this app for the iPhone that was like, oh, Open AI on the iPhone. And then literally two days later, the official open AI app came out and it's like, okay, well that was a waste of $30 and whoever created that app is now dead. So like chill.
Yeah, that makes a lot of sense. Okay, my question before we wrap up is for people who are listening who are like, okay, I feel like this might be something I want to implement in my business. Like what are things that they should be thinking about? I mean, security comes to mind. Privacy. I don't know what of these kind of issues people should be thinking about.
I would be quite worried about putting confidential stuff into open AI. Um, it's sort of murky whether that stuff will actually be used to train the model. They have some parts saying no, it won't. And another part saying, I don't know, um, you wouldn't really want your confidential data to be spat back out to someone else to complete their business bullshit lingo. And at the same time, you don't want to give another company this stuff in general, right? You should think about like what kind of data is actually sort of private to me. I mean I would think about this even in my personal life. listing things to open ai, you are literally putting them into a web form, right? You know that it's not like just going to a genie in the cloud, it's going to a database that'll save all that stuff. And what usually happens to databases of saved confidential stuff, eventually they either get hacked, they get leaked, or the employees of the company will have a peek.
So you just gotta be mindful of that, not fearful again, just like, all right, hey, this is a new thing, but this is still a commercial company. OpenAI started out at this, uh, altruistic nonprofit thing and now it's some weird, uh, hybrid beast that actually has a bunch of not just a bunch billions of investment dollars that want a fantastical return. And so do all these other AI businesses. If you look at where all the investment dollars have gone for the last six months in tech, like 99% of it appears to have gone into AI. All that money is gonna wanna return. Some of those people are gonna feel the squeeze. And when people feel the squeeze, ethics and morals usually bend. So just, you know, have that in mind.
I think the quote of the episode, it's not going to a genie in the sky, definition of AI.
It rhymes too.
I love it.
Did you ask Chat GPT?
I did just ask Chat GPT. Yeah, you did. Didn't come up with it, .
I love that. Uh, well we're gonna wrap up. If you like to watch your podcast instead of just listening to them, Rework is now on YouTube, so we'll link to that in the show notes so you can find it and watch in addition to listen. Rework is a production of 37signals. You can find show notes and transcripts on our website at 37signals.com/podcast. And as always, if you have a question for Jason or David about a better way to work or run your business, leave us a voicemail at 708-628-7850 and we just might answer your question on an upcoming show.