Rise of the Alexa Army: How 500 Million Devices Are Shaping Our Lives

0:28Welcome to the ChatGPT podcast. that go out to people, but for some reason, they're not responding back to those. So, but they do respond to when I text them, like with my iPhone number, I don't know if it's because they get like the blue text, or if my wording is different. But basically, to customize the text message for all of them, I put them into I put this message into prompt box, and I turn their name into a variable. And so as I go through all these leads, I'm able to copy and paste their phone number into a message. And then I just copy a prompt box thing. And it's like, what's their name i put their name it's like what competitor were they using i put that into and then i copy that so i only have to type two things get a whole message paste that into the text and then i send it to them they respond to that but they don't respond like all my auto stuff but yeah that's kind of a fun funny use case that i'm using prompt box for so was your was your auto response the exact same like formula of like words and stuff yeah maybe a little different because i'm lazy to go into like zapier and and change it all up but yeah like use their name and i don't think it used their competitor so i do have that in this new one um do you know if the the ones you were doing before if they had like um if it was a blue text like you're saying on iphone like yeah the old one you can't do it like i tried to find a way to automate it through my actual phone but all text messaging software like uses these like nine digit numbers that don't always have good deliverability.

1:57I'm trying to like right now, I applied for like a short code, which is like a six digit number. And you basically have like full on deliverability, but you have to kind of convince the text message marketing platform that you are doing transactional text, like, you know, like if you buy something and it's like, Oh, your purchase is verified or whatever. Right. Okay. Gotcha. Kind of trying to know we're not marketing. We're just sending people text messages. That's awesome. That's, that's really funny though. Cause I feel like, um, That brings me to like this whole concept around AI where I feel like if people feel like something's automated, if it's AI generated, like it definitely has a lower perceived value because you just know like the effort that went into it.

2:35So it just has lower perceived value. But if they see like, oh, this came from an actual iPhone, it's like a blue bubble on my phone, meaning another iPhone sent this to me. All of a sudden, like the perceived value is so much higher. And I think that's going to be a trend with AI where like anything you can do to make it seem a little bit more human or show people that there was some sort of human input. or touch will all of a sudden have like dollars attached to it because everything else can be just like 100 % automated. Exactly. Yeah. Like, uh, when I'm hiring for jobs now and like Upwork and Fiverr trying to find freelance freelancers before I used to get really badly written, like really like too formal, but like very badly written things.

3:16Now I get perfect grammar. Everything is perfect. I can tell they're using chat GPT to send me stuff, but it's like over the top and it's like to the point where it doesn't make sense they they still don't know how to like use prompts properly to like convince me that they're like competent so eventually like they're gonna find a prompt where it's like speaking like uh you know casual california bro voice or whatever you know they're good they're gonna find the magic sauce but like they right maybe they never will that's really funny well another interesting uh thing that i remember you telling me about and that I recently used and actually worked really good is when making like a job listing on something like Upwork to hire someone, you said, um, to make sure that you've actually read this, please start your message with the word like howdy Matt.

4:04Um, anyways, I used that and it was, it was really awesome. It was a quick way to like filter through, uh, what were just like auto response and what's what were like actual people. Cause they would be like, howdy Jaden, blah, blah, blah. And respond to my like job listing. That being said though, it did come to my like like i i realized that if you were using chat gpt and you just copied an entire job listing yeah and until it said jgp respond to this chat gpt would know like read it and say it would know to start it with howdy so i'm like ah pretty much i guess it just weeds out like really poor uh poor quality applicants that didn't read it and didn't use anything to like try to clean it up or make it more customized so i mean oh interesting like like you definitely are cutting out probably the lowest quality responses like zero yeah just like completely automated but it's not cutting out chat gpt so that is kind of an interesting uh yeah it's hard how would you do that like even if you ask them for like a specific plan on how they would i guess you could give them no context on what your business does and say go to this website and learn about it and and then tell me how you would you know like for example i if i want them to do like video editing i'll attach a file and I'll say like tell me how you would edit this up or or something like that something that requires like another step you know yeah yeah download this and send me like a sample of your work or something yeah that's true that's true one sort of verification would be like they would have to do any sort of file download be like start your message with the word in this file and then or start your message with like start the message with the name of my um you know the name of my like software project name and which is linked in the file picture and you have like you know the picture the screenshot of whatever you want them to work on so something big in there that they have to like actually click on and look at um dude did you make your own captcha for them you have to they have to open up they have to open a picture and say like this is a yellow duck and then they can reply to your job posting so you know they're verified like real people it's not just chat gpt oh man well you know the funny thing is there's like on a lot of these job things There's people that they'll have like one account and they almost like loan it out to people to like use their account.

6:12I know it drives me nuts because it's like, oh, they've done like a million dollars in sales. But it's really like seven guys all using the same account, just doing tons of different jobs on there. And so I know those kind of guys definitely just have software that just go out and like spam every single job posting saying, hey, I can do this for, you know, your price or cheaper. And then they just whatever they get, they just bring it in and offshore it to someone. Yeah, it makes me mad because they send their first message and say like the guy on the account is named like Muhammad. And he'll say, hey, he'll message me and be like, hey, this is Asim.

6:45I'm like, who the heck? What the heck? What the frick? Who are you? I just read all about Muhammad. Like I hate this. Oh, yeah. I lend it to Muhammad sometimes. I'm like, probably the other way around. But let's see what you can do. Okay. Well, I wanted to talk first and get your opinion. OpenAI has just launched a new ChatGPT feature, which is obviously exciting when that happens because it seems to be kind of the top. But before I get into that, have you had the chance to mess around with any of these other AI models that are coming out? Like obviously Google BARD and obviously Anthropics Cloud.

7:22Have you had the chance to mess around with either? I guess BARD, probably most important. Yeah, just with BARD. And BARD sucked. Really? I used it the first day. yeah it came out and i was like oh let's check it out i'm like i bet it's gonna be awesome and then i asked it like three questions i think i was just trying to get it to like say something naughty like classic like let's test let's test the uh abilities and then it gave me the answer to a question i didn't ask it gave me the score to a basketball game it was like oh the lakers scored like this many points in this season with like this person or something and i was like what like i asked you like about this the geography thing it wasn't even close and i looked it up and it turns out that uh bard will like mix up your questions with other people's from different accounts and i'm like uh i didn't ask any i'm like what is this answer and um it's like you asked a question you asked this question and it like phrased the question that it thought i asked I said, no, I didn't.

8:22I don't know if you've ever told Bard, like, no, you're wrong. But anytime I told it, no, you're wrong, it's like, yeah, I am wrong. And I was like, how far can I push it? Because I'm like, is it sky blue? It's like, yes. I'm like, no, it's not. It's like, you're right.

8:38It seems useless. It's like, yeah, an AI model that's like the customer is always right. Like somewhere in its prompt. That's hilarious. Yeah, yeah. Yeah, somewhere deep in its mind, it has to like give in. Yeah. And I've seen some funny stuff on Reddit and stuff. I don't know if you, have you used them both? Yeah, yeah. I've messed around with them both. You know, I think BARD has some, like a lot of cool potential. I mean, it's Google. So they have, they should have all the same data that OpenAI has because they've indexed every website on the internet. And at the same time, they have like some cool integrations, like essentially a lot of the plugins that chad gpt is integrating and launching um like google bard should have like they have google flights they know everything in everyone's email they know like they have like a lot of data that even opening i wouldn't have and so uh they know how people are using like a lot of corporate tools um with like their workspace i guess uh open it uh i guess uh opening i probably has that with their microsoft partnership but anyways i feel like they should have like a lot of extra data that would make them more powerful but at the same time yeah i keep feeling like They're still a little behind OpenAI on stuff, which doesn't feel good when they took an extra six months to further develop their product.

9:53Yeah, it's crazy. I just used, I don't know if on Gmail you have this. I was typing an email on my phone. Maybe you saw this on my Instagram story, but it's like I said, this is the last email I ever write because there's like a button. Literally, Gmail is writing my emails for me at this point. So have you used that a lot? is that um what's your like uh rating or review on the outputs it creates it's fine i've only used it on my phone but the problem is it still asks me like what the prompt is it's like what would you like this email to say but what i'd rather do is read the email in fact read all the emails and then let me know what it thinks i should say and maybe i can start training it on my business and who i am and then like it'd be cool if it wrote it for you and you're and it's like how accurate is it from one to ten i'm like ah this is like a six and it's like what can we do better Like, well, instead of using this word, or referencing that thing, like do this instead.

10:44And it would be cool if you could train your own. yeah i feel like we're we're getting to a phase where like everyone and i think there's some pros and consents but i feel like we're getting to a phase where everyone's going to have their own like custom tuned ai outputs based off of them their beliefs their ideologies the way they talk the way they say things like why would i want um you know why would i want an output from open ai that says like it doesn't believe in this certain thing or it doesn't think this certain thing is good or this thing is bad when it like goes against whatever i think right so i'm obviously not going to use that output so eventually these things are going to have to be whether open ai custom tunes them or someone else does eventually these things are going to have to be like customized for the individual i i see i feel like we get to a place where yeah trained off of your all of your um data on gmail all of your photos and your camera roll all of your content you've ever posted on youtube or wherever right facebook like these things train so it's like you but then my thought is like what happens when like someone wants to like turn over a new leaf like you know there's maybe there's this some guy and he's like a thug and it's trained to have like thug AI goes to jail comes back like five years later and is like I've turned over a new leap I want to be a new guy his personality has changed he's reformed and then like he's got his old AI model how does it like how would you know what I mean though or even I'm like that's a dumb such a dumb example but but but such a but no no so like I think it would more realistically be like this is the AI that controls like my business email, like my business stuff.

12:14But like, what about like personal email stuff? Look, I'm just looking out for you whenever you inevitably go to prison and you're going to get a reformed man. The guy's like, yo, like stop using the word like pimping so much in my emails, man.

12:33It ain't cool. Okay. But you get the idea like, okay, Okay, so work and personal. That's a good like one. I guess the way we talk to different people, the way you might talk to like your kids or your spouse or like a coworker would be different. Different people talk different, different social circumstances. But like what if you really wanted to like, I don't know, change some drastic thing. You want to become a new person. You want to like remove some old habit or introduce some new habit. Like how do you see these things being able to like evolve with you when the entire history of your life might be one way?

13:06Does that make it harder for people? I guess I'm coming from people probably laugh, but I'm coming from like a perspective of like personal development background, right? Like I've run self pause for a few years. It's all about like changing your mindset, changing your behaviors and like, uh, you know, reaching your goals, blah, blah, blah. But I feel like if you had like a mindset, perhaps like, uh, you didn't have a growth mindset, um, for example, and you wanted to like try to shift your mindset to be more like that, but your entire AI was trained one way. And so it just keeps spitting out like what you used to be.

13:34How hard is that going to be to change? What do you think the AIs do? Dude, I think that it's hard for people to change regardless. In fact, my thesis is that people don't change. Like 99 % of the time, like even if they want to, they don't. But I think that what I really want, because I've been wanting to make my own bot for sales and for my website. And I know you can like feed it documents. Yeah. But I think as long as there's like a, what I want is like, I want an AI box. i want an ai box where i can open up a prompt and just tell it something new and it will like adapt to that and be like hey instead of doing this like we always do do this instead because it's the only way you know even in my company as we do things differently at prompt box or app rabbit or you do things differently with like your self-pause coaching or anything like that it's like hey we don't do this anymore like we do this instead and i just wanted to like make all the changes retroactively and like just you know be new yeah i think that's the solution and i think i think your example there is probably a lot more viable than my like you know some person turns over a new leaf one but it's like we've done something we've always done something one way in our business and now we want to change how we're going to do that we want to change our company culture this is that's probably a lot more realistic right because people do go through like massive company culture overhauls or they try to they try to do like brand new initiatives in their company this is pretty common i feel like um and so yeah it's like how do we actually get the ai to do it and i think yeah maybe having some like model override almost like um when you're tuning a model and you have like the weights and stuff you have like some big ones that are like these are like heavy hitter like ultimate um things and when you go and put a new thing like okay we're changing how we do this to completely be like this instead it needs to be able to like uh pull that through on everything it does yeah i think that yeah another another way that i do uh i tell ai i've been using chat gpt to write me ads and i an email copy and i don't take the end result that it gives me i always fine tune it but i always ask it what questions do you need to know to write a three-piece email chain sequence or whatever for marketing for people that sign up for my product called app rabbit and then it'll ask me like 15 questions like and i take like half an hour answering like full like long sentences and like give it like and it's asking me questions like do you have any promotions coming up and like like what are the benefits of the product how do you differentiate from your competition and like what are some features that you feel like people would be most excited about and like what are some things in the pipeline like really everything and then i take what i've given it and i save it in the prompt box as a and i i call the the prompt everything about app rabbit or everything about whatever it is i'm telling it about and the next time.

16:18I know I can probably look up the conversations, but I never do. So I just, next time I'm in chat GPT, I just take that prompt. I just copy paste it. And I said, given the information I just gave you, uh, do this now and I can get it to do something else for my business. And I just gave it like, you know, 500 words of text about my business. And then I can just reuse that. Dude. I absolutely love what you said. I think, okay. Something in there that you said was an absolute nugget that I feel like absolutely you have to do, this is going to make every single prompt and every single AI output and any project you're doing in life better.

16:51You don't just go and tell ChatGPT exactly what you want because by doing that, you're limited to what you know in your mind. What you have to do and your current experience, what you have to do is ask it, what do I need in order to accomplish this outcome, right? Like what are the questions I need to ask you to get all of the right information in an email chain or all of the right content to an investor or all of the right information for a customer, ask it like, what do I need? Like, what do I need to include? It gives you the questions, you answer the questions, and then it gives you the output.

17:19I think that's like the absolute golden strategy to actually get really high quality content. Yeah. With everything, whether it be podcasting or content. I mean, isn't that the premise of Hitchhiker's Guide to the Galaxy? It's all about just asking, finding the right question to ask. I think that's why Elon is obsessed about it because he's always the right question. Yeah, exactly. So, okay. So the feature that I was talking about a while ago got sidetracked, I guess, is that ChatGPT, number one, Sam Altman just announced that he's allowing everyone to have plugins. I think after Bard came out and it was free, he's worried that, you know, everyone's going to see all the functionality and switch over there.

17:57So that got going. The other one that I think this is just funny. I love capitalism. I love capitalism. Yeah. The other one that I think is funny is he added a button that just says like continue, um which is hilarious because i think like right now um you know when it when it like caps out halfway through a response because it runs out of tokens or whatever you know i just think all the people aren't going to know the pain that all the ogs had to go through by typing please continue writing like this thing and it continued to write now they just got a button like continue generating or whatever that's pretty funny um i honestly surprised i didn't add it sooner since it's it probably is just a like a it's just a continuation of the prompt anyways it's a button but it just does what you type anyways.

18:37It's just probably the most common thing people message to it. So they turned into a button, but that's funny. Do you see chat GPT integrating a lot of complex settings, features like first example, something like a prompt box. Do you see them ever integrating something like that into just chat GPT itself? Do you think there's a reason why they've been like pretty light on any features other than literally the chat and like, the history. Yeah. I was actually looking at, um, I'm actually very interested in the plugins. In fact, some people have said that the plugins for chat GPT are like the largest opportunity in the world right now.

19:19And I kind of agree. Are you on the wait list? I am on the wait list. Uh, I have chat GPT four, so I don't even know. I should know the, uh, the wait list to create a plugin. Are you on that wait list? Yeah, I am. Uh, I need to check my email to see, but like, Even my chat GPT-4 doesn't have all the stuff that some people have yet. Yeah. I don't have internet yet. They just launched it. So I would say this for anyone listening that might be in your same place. Go to GPT-4. Go to settings. And then inside of settings, there's a toggle you can turn on to enable plugins. But I think they've rolled it out to everyone now.

19:59Oh, dude, you're right. It's in beta features. Yep, yep. They kind of buried it. And I'm not 100 % sure why. Like if this is just an awesome feature, why don't you just give it to everyone auto enabled? Like you go onto your account. I'm not sure why they have a toggle in the settings. I wonder if it's like, cause Sam Altman, he tweeted and he was like, Hey, we're enabling this for everyone. Maybe it's like their way to slow roll the launch. Cause not everyone's going to know. And they're gonna have to hear someone like, like, I didn't know until I saw someone tweet about it. And now you didn't know until you talked about it.

20:25And everyone on the podcast now knows, but like, I'm not sure why they wouldn't just roll it out because like Google's was just integrated and rolled out. And that's who they're worried about competing with anyways. So I'm not 100 % sure what their mentality for slow rolling that would be. They must be a bit sketched out by it or something. But yeah, the plugins. Oh, cool. So now I can select a plugin. That's, oh, the plugin store. See the, that's, that's going to be, so it says plugins are powered by third party applications that are not controlled by open AI. Be sure to, you trust a plugin, blah, blah, blah.

20:55I wonder how strict the store is going to be. If it's going to be like Apple, Apple and Google app stores, for those of you who don't know, insanely hard. like you'll probably get rejected at least twice before getting any app passed through them. I hope that one day they just completely die and everything turns into a web app. But with chat GPT doing this coming from a guy that has a company that creates apps for people. Yeah. Literally the, the coolest part of about app rabbit is that you can get your own custom app on the app store. And I hope that they die. It just, it just sucks. It also gives us a moat though.

21:27And honestly, I don't think it is going anywhere for a while, but yeah, I think, uh, uh, the plugins are like giant and to answer your question about prompt box i think that they if like i need to build a plugin for chat gpt for prompt box but you know part of me every month is like this is a month where prompt box gets like taken out like there's one that just got taken out i thought last night i'll have to look it up i'll look it up um in a second but it's like a plugin comes out and boom like a company just becomes irrelevant all of a sudden right what But what do you think about plugins and the kind of the landscape now?

22:06You know, I think that there is incredible power to first mover advantage. Definitely, I think if the big dog, so like if OpenAI releases a product, it's going to crush like a lot of competitors. But if it's just like other competitors or other companies coming out, it's like most people will not have heard about certain things. So I guess as long as you're competitive in like features, whoever comes out first is going to have like a really big advantage, I think. So yeah, for anyone developing software, the other thing is, I think if you just build a product and stop, you'll probably get it overtaken.

22:44But that's just, I feel like that's just business in general, especially in software. The secret is first mover advantage. Like, so if you're like, okay, no one's doing this yet, but I'm worried someone's gonna. Okay, well, do it first. Do it good. If someone else does it and you already, if you've already built up some sort of audience there, you already have customers there, you can pivot and change and like do extra things. Like every big company, like Intel completely shifted and pivoted to doing like their chips when there was some major changes in technology. And like a lot of other companies have done similar things.

23:11So like, I wouldn't be afraid like, oh, some company could come in and potentially crush it. Just get in at first, get a bigger audience, and then you can pivot and change. It's like the audience that's the power and having that audience. So I think a bias towards action will be good and I wouldn't worry about it. I mean, you can't worry about it. Like you can either do nothing or you can do something. So I would say do something and that's - You have to build and be loud. Yep. Yeah. Yeah, 100%. So many cool plugins. Like I'm looking at one, Calendar AI. Calendar AI sales agents generate revenue with potential customers from 200 million companies globally.

23:48World news. Summarize news headlines. You can ask for the latest news from various sources around the world. One I saw last night on Twitter was this called chat with PDF. Too busy to read a doc? Just upload it in chat GPT and chat with it. Ask questions and ask for insights. Yeah. Yeah. All super, super powerful. I feel like something like that. It's interesting because I don't know if like if you build a plugin and it becomes a really popular plugin like at what level does chat gpt try to like like why wouldn't they just integrate that like i'm sure google bard is going to integrate that you know what i mean yeah i think i don't know i i don't know i don't know because i don't know how heavy it is to lift the whole ai back end it might not be that crazy like i know they're have the algorithms and the ai learning and everything but maybe they're content to just be like the guys who you know if you want to use a plugin they're gatekeeping all of it right so right and i think like it's probably the concern everyone has building an ai right they're like is my project going to just get done by chat gpt or open ai and just integrated directly but i just think like i'm not sure what the what the relationship it is if it's like some sort of conflict of interest because if they do start integrating a lot of like features directly in it's kind of brutal if you built something let's say it was like the number one plugin and then chat gp is like sweet now they just like instantly like uh integrate it it's kind of like uh you know they kind of did them dirty because they they got all the the user feedback that this is a popular tool before they just did it yeah that would be a slap in the face and i think they probably want it to be like a pretty active development pool dude this is crazy there's no search option for plugins you literally have like a table like like a previous and a next and you have 10 plus pages of new plugins but there's no search bar so even if i even if i wanted to go find like one of these cool ones like there's a cool one where you can just paste a youtube link in and i already have a an extension called glasp that does this but you paste a youtube link and it gives you like bullet points of like what it's about but there's a extension which i'd rather use because it's just all in chat gpt but for me to find it i literally would have to scroll through 10 pages of plugins oh yeah that's yes that's the other thing i just feel like sometimes they're doing things like in a really not uh effective manner like they're really light on features like why wouldn't you include a search bar like some things that just seem so basic they just don't have and like you know we know that they have the budget for it so So it's just, it is kind of a little strange.

26:24Oh, I found a text messaging one called glowing. I'm definitely schedule and send daily SMS messages. Dude, if that can incorporate AI into the actual message and respond, like I'm all over it. I bet it doesn't, but that would be awesome for dude. Well, it may not have the blue bubbles, so it might be less effective than, but yeah, it would be cool. Okay, so two big stories came out in education. I want to get your opinion on them. The first one is that New York public schools have completely dropped their ban on ChatGPT, which was the first, you know, it was like one of the very first education platforms to do that.

27:03So what's your thoughts on that? Good. And also Italy dropped their ban, I think, too. I think their whole country banned it. Yeah. So I think it's good. I think, yeah, I mean, we had a school licensed prompt box so they could have like a white labeled version of it. so they can teach their kids how to use prompts. And he's like, I need it for all my students. And so we came to an agreement and that's pretty wild. I think it's awesome though, because if AI is the future, like if I was a teacher in school right now, I definitely want to be teaching the kids like how to use AI effectively. It kind of puts everyone back in the director's chair, right?

27:44Like you don't have to, maybe it's not good for your brain development to not have to like write the essays and stuff, but it is interesting to think why do i want to write the essay and like what do i want it to be about and how do i want people to see it and that's all you really need to be thinking about that's that's an interesting one it might not be healthy i don't know yeah so people some things some things i think about that um is we kind of had this initial wave where there was a bunch of like you know italy and like the new york school system and i felt like we're starting to see a bunch of people banning stuff and it was really left to be determined like is this a wave that kind of goes through where everyone's going to ban it and like try to shut it down or like, are we going to embrace it?

28:21And I think by Italy unbanning, New York school system unbanning, like I feel like we're getting to a point where the wave was very small and quickly reversed where people realize there's no way they can possibly stop this, nor is it going to be helpful to do it. But then another news story just recently came out where a Texas A &M professor, he threatened to essentially fail his entire class for using ChatGPT. And a lot of people are criticizing this. It's pretty funny because he said, hey, like I'm going to take or I took your last three reports that I made you guys all do and I put them into ChatGPT and I said, hey, ChatGPT, did you write this?

28:59And he said, it said it wrote every single one. So like all of you guys are pretty much failing and you can't graduate. So a couple like everyone's criticizing, obviously, because like chat gpt doesn't actually work like that you can't give it content and ask it if it wrote it so the guy is you know obviously way off because of that but just the concept of you know failing everyone for using ai i feel like goes to the same thing as the new york city ban um and is gonna inevitably just have to be uh is gonna have to be wiped like you're gonna have to learn how to use it as a tool um because there's no way people can completely get around it plus you can just use tools to make it so it's not detectable by AI anyways.

29:38Yeah, no, a hundred percent. It's, uh, it's like stopping people from using calculators in real life now. Like if you assign homework, really what they'll have to do is just not assign homework and have people write in class. Although I don't know if people are on laptops in school now. I was in a notebook, pen and paper. Some people had laptops, some, some, some of like the super nerds would bring like laptops and like they were allowed to write in class, which was probably more effective, honestly. But they're just gonna have to have like computers with no internet and people are gonna have to actually write, which is good because I actually never agreed with homework anyways.

30:15I think if you're going to school for that many hours a day, I don't even think you should go home and then do more school work. I think that should be the time where you should be doing other stuff, you know? Right, having a life. Okay, so here is a really big piece of AI news I'd like to get your feedback on. So essentially, Amazon just said that they're going to bring a chat GPT type AI to Alexa. And like, bear in mind, Alexa has sold over 500 million devices. Do you think this is going to, what do you think the impact on the field of AI is? This isn't necessarily like chat GPT, where people are sitting down and typing.

30:47It's very like voice based. Do you think that that's going to change consumers behavior to like use chat GPT less if they can just talk to their Alexa, for example? hmm i don't know i wonder if alexa will like suggest things to them like oh given the conversation you had earlier uh you know you might want to try this like this this low toxic like washing detergent or whatever like for clothes like it's gonna infer trained off of everything you've ever said it just listens right it's like hey amazon just came out with this like there's a new product called like this you know bamboo sheet sheet thing it's like maybe uh you know maybe you should give it a go in fact i'll order it for you right now all you have to do is say yes you're like uh yes well uh i think you bring up a good point because whether amazon alexa does this or someone else i absolutely guarantee someone's going to do this in fact i recently covered a product called humane which is like it's kind of like a little yeah it's a little camera and microchip thing and like recorder that goes in your like shirt pocket and it watches everything and you can like talk to it and stuff.

31:56Okay, whether it's Humane or whether it's Amazon Alexa, I can guarantee someone is going to build a product that is in your house listening to everything. And it's like, and it's not gonna be like a privacy concern. Like they're gonna shy away from that. They're gonna be like, hey, to like give you better accurate results, we listen to everything you say. So we really know what your tastes are, what you're interested in, what your opinions are. And we can really help make the most ultimate personalized experience for you. So in the demo that I saw of Humane, uh, it was the guy asked his little AI assistant that was in his pocket.

32:29If he was allowed to have like a donut or a bite of ice cream or something, did you see this? It's like a Ted talk. And it was like, yeah, like you have the calories to afford that today. Or like, as long as you don't do more than like two bites, like your, your dietary system can handle that. Or I forget what it said exactly, but I didn't, I didn't like the fact that he was like asking a robot, for permission like i feel like that's bad but it was interesting but did you hear afterwards it said do whatever you want it was it was really supportive of him yeah yeah this one said do whatever you want super nice yeah oh thank you robot in my pocket i can do it thank you for permission yeah thank you for granting me like my my rights right so anyways i'm really curious though because i think this is actually bigger news for a lot of reasons like obviously okay google's in the space now microsoft and open ai are in the space we got anthropic and claude which are like some indie projects which are kind of cool but to be fair um anthropic is does have like a two or three and i think it's a 300 million dollar investment from google so well you see like open ai aligning with microsoft anthropics kind of aligning with google um now you see like amazon coming out with one it's going to be like what apple next and um i don't know who maybe intel you see like ai products coming out of nvidia obviously they're making all the chips though so maybe they'll stay away from like the ai uh what do you think about the what do you think about the chat stuff like talking to it because that's like jarvis from iron man or it's it's like talking to robots has always been part of sci-fi that seems pretty realistic now especially once it gets into like robots and stuff like that i think it gets a whole lot better i actually think this is like uh like alexa is like the perfect use case for chat gpt because um in the past if you've used alexa or i had like a google play or like a what was it like a google home mini i think or something anyways it's an alexa from google and um eventually like like when you ask it for responses they had to like program it on how to crawl the internet to get the right response that you were looking for and i imagine that was pretty tricky where now you literally just plug it into an API or you build your own tech on Amazon's, you know, side where you plug it into like your large language model.