Open AI CEO SHOCKS Everybody About GPT-5 (GPT-5 Update)

For those of you who are wondering exactly what's going to happen with GPT5, this video is going to tell you quite a lot because The CEO of OpenAI actually had some very interesting information to say in a new recent interview. So check out this article right here. You can see that the CEO has confirmed that GPT5 is not in training and won't be for some time. So let's take a look at the clip which surfaced online in which he details the exact details of how they are training GPT4 and, and exactly when we can expect GPT5 and if they are going to be trading this model anytime soon.

Like most technical nuance about what's where we need the pause, like it's actually like OpenAI, an earlier version of the letter claimed OpenAI is trained in GPT5. Right now we are not in the world for some time, so in that sense it was sort of silly. But we are doing other things on top of GPT4 that I think have all sorts of safety issues that are important to address and we're totally left out of the ladder. So I think moving with caution and an increasing rigor for safety issues is really important. The letter, I don't think is the optimal way to address it all.

See that the OpenAI CEO Sam Altman clearly discusses the role that GPT5 will play in the next level of AI evolution. And he discusses that he needs to make sure that this AI is actually going to be released soon, safely. And that is of course, something that was brought up in this recent letter. Paul's giant AI experiments. An Open Letter. We call on all labs to immediately pause for at least six months before training of AI systems more powerful than GPT4. You can see as of recording this video, it has around 27,000 signatures. The reason this has actually drawn up the attention of many world leaders and policymakers is because many influential people, such as Elon Musk have, have decided to sign this letter and publicly state that GPT5 is likely to be somewhat dangerous if there aren't certain protocols in place before it is released. Hence the reason for this pause in AI. So of course we now know that Sam Altman has responded to this by stating that GPT5 isn't even being trained at the moment and won't be for some time. Now, of course we know that in the letter it states that we call on all AI labs to immediately pause for at least six months to training of AI systems more powerful than GPT4. But you have to understand that many People haven't even caught up to GPT4 and GPT4 hasn't been fully rolled out yet. We're still missing many features such as the larger token size window and of course the visual multimodal feature which was teased in the upcoming trailer. So there's many different things that are to be discussed and there is something really interesting coming from Google that definitely impacts how GPT5 is going to be trained that I need to discuss later later on in the video. Now, of course, Sam Altman did state some more stuff about GPT5 and the letter on Twitter. He stated right here that one thing up in the debate about the Paul's letter I really agree with is that OpenAI should make great alignment data set and alignment evaluations and then release those. So essentially what he's saying here is that whichever alignment structure that OpenAI use, they should definitely release that because it's going to be something that other companies need to use as well in order to make sure that AI is going to be released safely. As AI safety is definitely one of those issues that if you don't fix it, it's going to be very hard to put the genie back in the bottle, as some would put it. Then of course we had Greg Brockman talking a lot about GPT4 and GPT5 in terms of how training times are actually done. So he said the underlying spirit in many debates about the pace of AI progress is that we need to take safety very seriously and proceed with caution. It's our key mission. We spent more than six months testing GPT4 and making it even safer and built it on years of alignment research that we pursued in anticipation of models like GPT4. So essentially what he's saying is that he knows that when they release newer AI tools that of course safety is one of the biggest priorities because they don't know how powerful this technology is going to be. Because a large amount of the times there are these things called emergent abilities where essentially when an AI gets smart enough, certain abilities just become present because of the large scale nature of that large language model. And these abilities aren't taught, they simply just emerge. And of course if you have abilities that do emerge, you're going to have to ensure that there is a long rigorous process in which you ensure that the AI is completely safe before it is released into softwares like ChatGPT or many different API tools which use the open APIs large language models to run their different businesses. If you were confused by my description just there, take a look at this definition, an emergent ability is a characteristic or skill that arises spontaneously from the interactions and complexities within a system rather than being explicitly programmed or designed into it. Now, I watched two videos that were very interesting about the dangers of this, and I think you should take a look at a few clips because if you understand exactly the examples that we've been seeing, you understand why a pause on GPT5 may be necessary to understand and make sure that These models are 100% completely safe.

These models have capabilities. We do not understand how they show up, when they show up, or why they show up. Again, not something that you would say of like the old class of AI. So here's an example. These are two different models, GPT and then a different model by Google. And, and there's no difference in the, the models, they just increase in parameter size. That is, they just, they just get bigger. You ask the these AIs to do arithmetic and they can't do them. They can't do them and they can't do them and at some point, boom, they just gain the ability to do arithmetic. No one can actually predict when that'll happen. Here's another example which is, you know, you train these models on all of the Internet, so it's seen many different languages, but then you only train them to answer questions in English. So it's learned how to answer questions in English. But you increase the model size, you increase the model size and at some point, boom, it starts being able to do question and answers in Persian. No one knows why.

So right there you can see that this is definitely something that puts the alarm bell for creators who are making AI and even those who are trying to regulate AI. Because imagine making GPT5 and it's able to do certain things that we don't understand. And remember the some of these abilities we don't yet realize until maybe months or even years after. For example, there's theory of mind, which we only realized that GPT4 had. And then when we looked back, we realized that there was some theory of mind in GPT3 and in GPT2. Now this is also prevalent in other AI software. In a recent interview where Google's CEO was talking with 60 Minutes, they detail how it was able to learn an ability which it didn't previously have coded into it. So I can't play the clip due to copyright, but I will have a link to the interview in the description of this video. So I do think that that is a very, very important point that has been brought up by Many of the AI researchers who are looking to regulate this software because of course, as we do know, we're kind of stepping into the unknown as the technology kind of just rapidly, rapidly advances. But of course, with regards to GPT4, there are still other companies like Google who are working on trying to beat ChatGPT and GPT4, but by pumping billions into many smaller companies that have large language models with them. For example, they are pumping millions into a bot that's called Claude Next, which is very, very good and quite close to ChatGPT. And their goal is literally to create something, get this, that is 10 times more powerful than GPT4 within 18 months. So that is definitely a very, very quick ramp up if that is their final target. Which means that we could definitely be seeing the AI race heat up even more than we even thought, and having these companies rush out software which sometimes might not even be ready, just so that they can potentially own the largest part of the future. Now, Sam Altman also discusses right here things about a good AGI future. Number one, the technical ability to align a super intelligence. And this is definitely something that we must focus on because if something is super intelligent, it is likely to have goals beyond our understanding and, and it's probably going to be one of those pivotal moments in the future if we don't get it right. And of course he also talks about an effective global regulatory framework, including democratic governance, where essentially you just have this framework that everyone adheres to because it's the only way to ensure that we do have a safe AI. So let me know your thoughts on this. Do you think GPT5 is going to be coming quicker once OpenAI realized that Google are starting to invest more in other companies to outpace them? Or, or are we going to be seeing a slowdown in the release of AI products? Now, something that I did forget to mention, and this is a very large point, is that GPT5 is going to be released incrementally, meaning that there's going to be subsequent updates every single month or every single couple of months that will lead to the eventual version of GPT5. Now the reason they did this is because they want to address the safety concerns and continuously enhance the AI model rather than one big rollout which could present a number of different issues. So definitely a really nice change.