ChatGPT Projects allows us to tidy up our endless sidebar of chats, and more importantly give context around the different things that we work on with ChatGPT, whether it's work or personal projects, we can upload files, manuals, reports etc to give knowledge context. But, don't forget to provide examples too, this is called 'tuning' the model, to guide it in its output, and get better results from the AI.#ai #aitools #artificialintelligence #chatgpt #chatgpt4 #chatgpttips #llm #techtok #tech #learnontiktok #learnwithtiktok #generativeai #genai

ChatGPT now has projects, but how should you use them? How is it different from custom GPTs? How is it different from Claude projects, which has been out a while? So firstly there are two reasons why you will use projects. The first one, the clue is in the icon, it's a folder, it's just simply a way for you to sort your ChatGPT conversations in much the same way that you sort emails into folders and other things into folders, right? So everything I ask about a particular client's project or my son's Roblox game that he's building are in one easy-to-find place, and that's a great feature and it's way better than the endless sidebar of chats. The second reason is the fact that you've got these conversations, files and custom instructions all in one place, so there's a very clear context and a knowledge base for every chat that you have there. Now to be clear, the custom instructions in a project take priority over the custom instructions in your account settings, but it still draws on those and it also still draws on its memory of you as well, also in your account settings. But yes, in a project it's the custom instructions and the files that it really looks at when it's responding to you, so you don't have to keep explaining the same stuff over and over again when you start a new chat, and that will save us a lot of time. So yes, on the more basic side I'm just gonna tidy up my sidebar, but the real power of this comes when you start giving it files containing really specific knowledge about the thing that you're working on. But one massive tip I would give you here is around something called tuning, which I will do a full video about, but we're finding that whilst files containing knowledge provide really specific context and give ammunition for better responses, tuning these models, i.e. giving them examples of the ideal response that you want, generally produces better results. So just think about that when you're uploading files here or to any AI tool, give examples as well as just knowledge. So that could be examples of tone of voice or style if you're using it for writing or content creation. If you're getting it to produce documentation in specific formats, then obviously examples of that just really helps it. So how is this different from custom GPTs you might be wondering? Well there are two key differences. A custom GPT is possibly a very early version of whatever this agent product is going to look like from OpenAI. By the way, agents is something we're all very excited about, and it's the idea that it'd be like having chat GPT that can go and perform multi-step tasks on your behalf to get a job done for you. That's possibly something we're going to see in the next year. So in custom GPTs now, you know I've got one that writes LinkedIn posts and it's trained on hook templates and successful post templates and examples, but it's hooked up to Google Sheets via Zapier so that when it creates content it adds those ideas as new rows in a specific Google Sheet. Okay so very very basic version of an agent essentially. And I could by the same mechanism have it draft emails, send slack messages, basically trigger any other software that I use. But projects don't have that. Secondly, you can share custom GPTs publicly and people can use them, whereas chat GPT projects don't have any kind of collaboration functionality at all. There's just no way you can share or collaborate. And that brings me to the differences between chat GPT projects and Claude projects. So Claude projects can be collaborative and you can also connect Google Docs, something you can't do in chat GPT projects. Anyway have a think about how this might focus how you use chat GPT. Think of it as different desks that you go to work on different stuff at, like research into a certain topic, content creation for a specific business, all your cooking recipes, personal finance planning, career goals and planning. An exercise plan that adapts to your results over time, right? It'll be really interesting to see what people do with it and I'll share the best use cases that I find.