← AI BLOG

GPT-4 is here. Where does this end?

JOHNNY MANDIOUS5 MIN READ
OpenAI ChatGPT announcement on a screen

From my archive, written 15 March 2023. These are my impressions at the time.

ChatGPT 4 was released, having many advantages to GPT 3. Firstly, its more intelligent in different languages, performs a lot better on complex exams. It understand images, not just what is shown, but it can actually understand any issues in an image and explain any humour present. There’s been improvements on coding to the point where you can send it a paper sketch of a website you want and it can give you the code to make the entire thing. It keeps a larger mental record of conversation; 64,000 words compared to GPT 3s 8,000 words and overall its more powerful, amazing the work that OpenAI are doing, next level shit, maybe even historical, it certainly looks that way right now.

Supposedly, Bing has updated their chat feature with GPT 4, but I can’t find out myself because I’m still banned from when I tried to jailbreak it, I don’t care too much though, since I mainly use GPT 3 anyway, I think once there’s been more testing on it, I should be unbanned, I did enter through a waiting list after all and its still very early on so they’re probably concerned about someone accessing some hidden files (which already happened, someone found a file explaining the early development of bing chat, which was originally called Sydney) Anyway, companies such as Duolingo, stripe and Khan academy are going to be using GPT 4, I only know of the example of Duolingo where they’re using it to make a bot that talks to you in a different language and explains where you went wrong, I tried to have a conversation in Spanish with GPT 3 and I think I’d benefit more from a Duolingo version because Duolingo understand the words I’ve already learnt and can generate text based on that instead of giving me stuff that I haven’t learnt yet and unlike with GPT 3, they’ll hopefully be able to give me feedback without ending the whole conversation.

I’ve heard bits and pieces about it, not sure about when its released, but if the release goes well and its not too much, I might consider buying the subscription for it, as fluent conversation is something I need help with. I’m just wondering where everyone else is, Apple? Google? Microsoft is smart for getting such a head start on this new and seemingly out of nowhere new technology. Apple could really really benefit using AI to upgrade Siri, in fact, just get rid of Siri at this point and make something new from the ground up.

Siri is one of the most useless features on an iPhone, maybe it was cool in its early release; a new cool assistant. But no person under 50 uses that shit anymore, they don’t need to, it has no use. In the last 5 years, I’ve only ever activated it by mistake, it doesn’t answer questions, it just searches the web for stuff, which would take 30 seconds more if you did it yourself, I think that it should either be upgraded or removed, but with new AI stuff like this, maybe they could make a better Siri that can do what GPT can, that would be useful, it would be more accessible too rather than having to join waiting lists or go on websites (since there’s not app for GPT 3/4) I’ve been tempted to buy GPT 4, but its £200 a year, a whole month of living expenses, so I’m not going to be able to.

I don’t blame them at all for making people pay, its something that’s losing them money and is something definitely worth paying for, GPT 3 was alone, let alone this. It’s weird how quick this has all gone, how in a few months, we have something way more intelligent than I ever expected would come so soon. Before any GPT and image generation, to me at least, AI seemed like an abstract concept, something that would one day exist but for now, isn’t a thing. If you told me a few years ago what would exist now, even in an early form, I think I would have expected it to be something that could think on its own and maybe we’re yet to see that day, but for now, all it is (as I understand it) is just a large collection of information.

It’s just another way that we’re able to obtain information and understand whatever it is we want to learn about. First though other people, then books, then the internet and now I think AI will be a new and better way that we can learn. It’s gone far beyond just presenting information and giving specific details about it, (even more than what the internet is able to find) but you can now create previously non-existent content; images, books, screenplays, coding, graphs, voices, accurate translation and even videos (although early tech exists for this) all in the span of about a year.

I think it will destroy some industries, maybe even a lot depending on how much its regulated, but I think its also a very powerful tool that may benefit individuals and businesses if used correctly. But I have to wonder where it ends, how smart its able to get until we can’t help but believe its human.