GPT-4 vs GPT-3.5 - First Impression! (Unraveling Minified React Code)

--- type: transcript date: 2023-03-14 youtube_id: 0OCg2UvBVDU --- # Transcript: GPT-4 vs GPT-3.5 - First Impression! (Unraveling Minified React Code) all right so in this one I just wanted to do a first impressions of gpt4 so if you have the premium subscription you'll be able to access it from the GUI right now there is a wait list for the API so I am on that wait list if you're interested in seeing videos on how to integrate open AI services with node.js there's a number of videos in my channel and I'll be making more content with chat GPT 4 or gpd4 in the API specifically so look for those if you're interested but in this one I just wanted to demonstrate it a little bit and show the things that I've noticed so in the interface you'll see that there is the drop down similar to what you get in the premium subscription where you had access to the Legacy model as well as the default model I think most of us have been using the default model where it is that turbo version where everything is just a lot faster and that sort of dovetails into my first impression of gpd4 and I think in part why they might have put these little bar charts on the left hand side here showing things like reasoning speed and conciseness because as soon as you put in a message into the prompt bar here for gpt4 you will notice it's considerably slower after using the default model that turbo model for for some time so I feel like that might be part of the reasoning is you're used to having that turbo version and really having those prompts come talk to you pretty quickly if you're using something like GitHub copilot that it might even feel even slower to use something like this but I think the thing with gpd4 to consider is that it does handle reasoning and conciseness considerably better so I'm still thinking of ideas on how to demonstrate this by all means put a prompt that you're curious in seeing um I'll aggregate a few of these in an upcoming video and demonstrate how it compares to the Legacy and default model here and the first thing that I just wanted to try was okay when I saw could handle the context of a lot more content is what will happen if I ask for a description of the react production minified library now this looks like just like a jumble of code to me as minified code does look like and I just wanted to demonstrate what it looks like between the turbo model and then gpt4 so if I say what is this code doing and I paste it in here it accepts all that code which is great and if I do it over here as well what is this code doing okay so we have to wait for the other message to finish and the main thing I just want to see is okay how well does it describe it so it's it's reasoning is apparently better it's conciseness is better so once this finishes I'll put a prompt here on the turbo model and we'll see what the difference is is it that much better to wait for the time or are a lot of people still going to be leveraging the GPT 3.5 turbo model Intel gpt4 speeds up a bit let me know your thoughts leave a comment below I'm I'm really just curious on exploring this um like everyone else right now so let's ask the same question of what is this code doing okay so we have a brief explanation and how about I say explain the code so as you can see here so the first impression is so it's trying to explain the code but it's it's explaining in the context of those minified variables whereas with gpt4 it was able to take that code and the minified variables and sort of abstract that out to how you would actually use that as a user so right off the bat it's that's a pretty interesting use case for just exploring this now obviously there's a lot of different contexts of how you can explore this I'm really curious how how others will be be using this and yeah if you have any ideas it doesn't have to be coding related but if it is that is even better give me an idea of what you want to see in comparison with gpt4 and gpt3 uh 3.5 and whatnot and until the next one