OpenAI has just announced that they're going to be making ChatGPT available generally to everyone without even needing to essentially create an account. So why is this a story and why is this interesting? A lot of people are talking about this. I made a whole blog post about this.
And this is interesting for a number of reasons. But first, exactly what this means is essentially, like you imagine, you go to the ChatGPT homepage and instead of being prompted to sign up or create an account, you can just go and start using ChatGPT, asking questions, etc. Now, of course, this isn't GPT-4, so it's still going to just be, you know, 3.5, their free version. But what's interesting here is it obviously costs money to, you know, generate.
So why are they giving this away for free? Now, the story answer is they're already doing this, right? 3.5, GPT-3.5 is already free to use. I think one of the big areas is they're just trying to reduce friction, essentially make this easier.
There's going to be some sort of percentage of people that are going to land on their homepage and are just going to, you know, bounce because they don't want to make an account. I tend to think, like, you'd be surprised by what that number is, by the way, because having apps myself, you know, SelfPause, the number one AI like coach that I've created, it is impressive the bounce rate you get on people that download the app and if you immediately make them sign in for an account versus if you just let them start using the app right away. And there's definitely trade-offs between the two of them, but it's interesting. I think it's significant.
I think they're going to get a big uptick. Now, what I think is actually interesting here is something that not a lot of people are talking about. Reddit recently did something very similar to this. And if you look at Reddit's web traffic right before, so Reddit was doing this pre-IPO, right, they're about to do a big IPO.
And I think essentially if you want to juice your numbers, this is one of the easiest ways to do this. Now, I don't know if that has anything to do with opening eyes strategy, but essentially there's a large percentage of people that, Reddit's, I guess, live different because there's content all over the web and it ranks very well on Google. But essentially what's happening is on Reddit a lot of times you're clicking on stuff and you're kind of immediately before you're able to read the thread getting hit with, like, make an account. Now, they stopped pretty flatlined web traffic and after they made the change were on Reddit, you didn't need an account anymore, you could immediately get your content, they saw a huge spike.
And it's in fact, like, noticeable enough that on their, you know, IPO filings, this was a big thing, a lot of people were talking about this, like, I think it was up like 40% or something crazy over a number of months, Reddit's, you know, traffic increasing because of this. And so I think while opening eye may not have, you know, necessarily content, I think this could definitely affect shared posts or shared threads. So, like, a lot of times you can actually share a thread that you've created, a conversation you've created, and that shareable link, being able to view that without actually having to log an account, I think that would be a benefit. I'm not sure if that's going to be part of the rollout or not.
But regardless, just being able to hit there and start searching, what people are saying is essentially this is going to turn into Google. And some people are saying with even better results and responses. So a reporter recently over at TechCrunch was asking some questions about this because apparently with this version in particular, they're going to have, quote, unquote, slightly more restrictive content policies. So the reporter asked them what that meant.
And this is the response he gave. They said, the sign-out experience will benefit from the existing safety migrations that are already built in the model, such as refusing to generate harmful content. In addition to these existing mitigations, we are also implementing additional safeguards specifically designed to address other forms of content that may be inappropriate for sign-out experience. We consider the potential ways in which a logged-out service could be used in inappropriate ways, informed by our understanding of capabilities of GPT-3-5 and risk assessments that we've been completed.
Okay, honestly, to me that's like a big nothing. It doesn't really explain. They've essentially said there's going to be some differences, but they haven't really said what they are. Your data being shared into the model, you can go and disable it, you go to settings, and then you go over to, there's a toggle that says improve the model for everyone, and essentially allows your content to be used to train the model.
You can toggle that off. I think a part of this is they're like, well, we get more users, but we're making everyone automatically opted into, we can use all of their conversations and training data. And so I think this is interesting. I'm not sure, like, when you look at Google, if they're, like, in my opinion, Google's probably taking all that exact same information for everyone that uses Google for free.
But, you know, people kind of complain about this when they have to pay for it, and so if it's free, it kind of makes sense. So then the one other thing I wanted to bring up is that same reporter, when they kind of asked about, like, what the slightly more restrictive content policies were going to be, they kind of gave that answer. They weren't really sure what that meant exactly, so they asked again, you know, and they got this response as well. They said, we've also carefully considered how we can detect, stop misuse of the signed-out experience and the team's responsibility for detecting, preventing, and responding to abuse have been involved through the design and implementation of this experience and will continue to inform its design moving forward.
I think the big thing here is people are like, what are you going to do when, if you don't even need an account, people are just going to create bots that go and, like, use this and send out answers and do a bunch of crazy stuff. The nice thing about accounts is, like, if there's an account that's, you know, doing something crazy and spending, getting millions of things generated, you know, okay, well, this account that's doing it, shut down the account. But if it's just open to everyone, you can have different IP addresses, different computers, different, you know, essentially different virtual private networks or servers all over the world, and one person can be controlling it to get tons of stuff. So it looks like ChaiChapin OpenAI is looking at this, but I'm curious to see how this will go.
Specifically, on their LinkedIn post, there were a couple of interesting comments. Someone said, please release the memory feature. Yes, this is something, this is, this is notorious. OpenAI announces something and they don't have it.
Roll out right away. They announced a memory feature a while back, and I'm not down there. Someone said, I'm curious on how you're protecting from an IP farm kind of scenario. This is what I was just mentioning.
I understand the cost of inference for 2.5 has come down enough. This is now a reality, but I'm still not convinced this can fully be protected against attacks for the team, protecting this from malicious actors. So, yeah, it's going to be interesting to see how that rolls out. Someone else said, removing friction is probably the most important thing that sets companies with amazing products from mediocre ones apart.
So I agree, like we mentioned with Reddit and now this move. I think removing this friction, they're going to see a big bump in site usage and traffic. Not big, but, like, I think double-digit percentage. And because of that, I think, whether that means that they're, like, going to look at raising money or they need that statistic for some other benchmark, I think that it's an interesting play to make.
Someone else said, no sign-up means easier access, but without an account, you're going to miss out on having, number one, the ability to share conversations, number two, conversation history for records retention, and number three, opting out of using data to train the model. Okay, so I guess you may not actually be able to opt out of using your data to train the model with this free version, which, if that's the case, that makes perfect sense whether you're doing that. She then said, all the more reason for organizations who don't have generative AI policy to get one in place. Yeah, that's a good point, right?
Like people are saying, you know, they're banning ChatGP at work early on, and then eventually they're, you know, you can opt out so it wasn't trained enough of any of your stuff if you paid for it. And so then they're like, okay, you know, organizations can use this. And, yeah, now you definitely want to have an AI policy because you don't want your employees to be using this free, not logged in version that definitely is taking all of the data. So a lot of very interesting stuff here.
Personally, I think I'm interested to see if they're going to be making any big announcements, if they're going to be raising any funding, if they're going to be doing anything in kind of that scope, what the need really to bump the numbers, or maybe it's just if they're opting everyone in automatically to have their data trained, they just want to get more data out of it, increase their users. Very, very interesting. Last thing I will say is in their blog post, they kind of started this off with a very interesting take that I think might directly be in relation to Elon Musk suing them for going from an open nonprofit to a closed for profit. They said, it's core mission to make tools like ChatGPT broadly available so people can experience the benefits of AI.
Okay, that is literally from like the lawsuit where Elon's like, you said it was going to be open source, and they're like, no, we didn't say it was open source, it's just going to be openly available to everyone to get broad benefits. They kind of changed that in their charter, I think. So anyways, it's kind of funny, they're like leading with this. They said more than 100 million people across 185 countries use ChatGPT weekly.
This is one thing I do want to say. They made this claim 100 million weekly, I want to say like six months ago, if I'm not mistaken. I feel like six months ago. Maybe time flies.
A demo day, I think, right? So maybe that was, what, November? So we got, yeah, a handful of months ago, and it doesn't seem like this has necessarily increased a lot since then. So I'm wondering if they're seeing growth plateau and whenever they make their next big announcement, they want to say 200 million or 140 or 150 million.
And so this is like trying to juice the numbers again. So, I mean, that's just one point for the juicing the numbers theory. They then say, to learn something new, find creative inspiration and get answers to the questions starting today, you can use it. Okay, whatever.
I think we got the juice out of that sentence there. In any case, I'll keep you up to date on everything that's happening with this. If we do see a significant numbers boost in the future, just remember this podcast episode and remember this may be what they're going for all along. Hope that you all have an amazing rest of your day.
Make sure to subscribe if you're on YouTube. Follow us if you're on Spotify or Apple. And make sure to leave us a review or a comment. I really, really appreciate it.
Hope that you all have an amazing rest of your day.