OpenAI's new ChatGPT Tasks feature schedules automated reminders and content deliveries on web and mobile. You can set daily gym reminders or weekly S&P 500 updates, pause tasks anytime, and get push or email notifications. The presenter calls it rudimentary but interesting, and sees it as a step toward agents that could eventually book travel or reorder groceries.
Chapters
OpenAI just released something pretty cool, which is ChatGPT tasks, which allows ChatGPT to go beyond chatting and create tasks for you. So think reminders, daily newsletters, daily stock updates, and so on and so forth. It's essentially like Google Tasks, but with a brain. It's not perfect, but it's pretty interesting. So I figured we'd show it here on the channel. And while you're here and before we begin, don't forget to like and subscribe to the channel. It really helps us and let us know what type of content you want us to create. Your opinions really matter to us. So with that said, let's get started. So I'm here on the announcement blog post where they discuss scheduled tasks and ChatGPT. As you can see, it's now rolling in beta to Plus Pro and Team Plan users and they're planning on expanding access when this goes live. So scheduled tasks are supported on ChatGPT, Web, iOS, Android, Mac OS. So everywhere you have ChatGPT, except for the Windows app and this will come later in Q1. And essentially as they described here, and As I mentioned, ChatGPT Tasks is just a way to let ChatGPT schedule tasks and reminders for you. We're gonna see that in action, right? And you can trigger them at a specific interval. So every day, every week or a one off schedule. And they will execute it regardless whether you're online or not, which is also interesting here. And you will get notifications, whether push notifications or email. Now some of the examples they mentioned here is, you know, give me a briefing on AI news every day, which is something that's very relevant for me, for example, or practice French with me daily, which is also really interesting. It's a learning use case. So they show you here to set up a task, but I'm going to show you directly myself. So I have here ChatGPT open. All you need to do is open ChatGPT and then under model, go to GPT4.0 with, with scheduled Tasks and beta and you'll be able to jump in. What's interesting is that the model name is called gpt4o jawbone, which is quite interesting. And an example here is of a task set, a reminder for every day at 8pm to go to the gym and it's going to create the task now. And once it's created, I'll be able to jump on it, edit it. So go to the gym, tell me to go to the gym, I'll be able to change the schedule for something daily, weekly, Monthly, annually, Custom, right. I'm going to keep it daily at 8pm I'll be able to pause it if I want or delete it. And that's about it. Now if I want to check out my existing tasks, all I need to do is press on my notif my profile picture here on the top right, go under here to Tasks and then I'll be able to see all the tasks that I have. I've already created a task on providing stock updates. I'll show it to you in a bit how that looks like in action. But that's essentially how to approach creating a task. Now an interesting example I mentioned here, the S&P 500 I'll show you how tasks looks like. Once you get a reminder or once you get the output from the tasks is essentially something that looks like this. So I just got via my email, as you can see here, an S&P 500 Growth and Risks Update email. I pressed on this link and I got this chat. So essentially it is a chart with the S&P 500 where I'm able to change here the different date filters, which is pretty nifty and with a discussion and kind of an Overview by the ChatGPT interface. Now with that said, if I want to update my notifications, I'll be able to go to settings, go on notifications and then for tasks go to push and email notifications. So I just showed you what email notifications look like. What push notifications look like will look something like this. So you're going to get a desktop notification. It will appear on your browser and you'll be able to jump on the task reminder. Now I just showed you two use cases here of a reminder to go to the gym at 8pm or downloading the stock prices or the S&P 500 stock every day. But I'll show you other useful use cases that I've seen from the community. So this is Karen Nguyen, she's an engineering, she's an engineer at OpenAI and here she mentions how you can delegate ChatGPT to write new Sci Fi stories in Canvas for you and send them to you every day. So if you're in the mood of reading an AI generated Sci Fi story every day, not sure why you would want that. You can get that every day in your inbox. Another interesting example here we said learning languages. So a daily new word or a daily new sentence, or even practicing with ChatGPT every day. It's a really interesting learning use case that could be useful here. And four, which is also an interesting data science or developer use case is if you have an interview coming up, you can practice a programming interview every day by being provided from ChatGPT a new challenge. And that's pretty interesting. Now it's important to note that these types of tools are only as useful as how you plan on using them. So some of the ways I'm Planning on using ChatGPT tasks is is ways to help out with my daily workflows. So for example, I want to be able to set reminders to get the latest AI news and recommendations for what type of content to create. And to be able to do that, all I need to do is go to schedule tasks again, set a reminder for every day at 8am where you send a summary of the latest AI news and five video recommendations or five recommendations for videos to create on these topics. So as you can see, it's creating the task and then it's going to send me that reminder every day and it's not going to be able to automate or augment that workflow, but it will provide some form of inspiration on top of my existing research workflow. Now, if you're learning data science or in the beginning of learning data science and AI, one thing that you can use this for is to send you new data sets every day that you can practice on or if you are, you know, interviewing for a data science role, as we saw here, using this to send you a new problem to practice on and sharpen your skills. There's many ways that you can use these types of tools. For now, in my opinion, this is quite rudimentary at the moment, is just a reminder used tool that is infused with some form of intelligence and is able to summarize information and provide more value than a typical reminder tool. That said, I do see this in the future becoming part of OpenAI's agent strategy. So Sam Altman and the OpenAI team numerous times have talked about the goal of building AGI or the goal of building agents. Um, and what that means is that these LLMs will go beyond just being able to provide you an answer to a query or a small close ended question, but they'll be able to perform complex open ended tasks. And I think this is one of the components that you need to be able to build an agent in the future. So not only here are they providing an LLM with the ability to plan future tasks, but if you project in the future, the ability to connect to different APIs, different data, different type of information that you have on your own computer for example you can expect some pretty wild use cases. So for example a travel planner that not only schedules reminders for you on where to go throughout your vacation but can potentially buy and book tickets and hotels for you. Or for example a grocery planner so a tool that not only is able to schedule reminders for you on when to buy groceries but that's also able to act on buying groceries for you and is able to replenish your fridge based on your consumption data or habits for example. So I do see tasks being part of a larger strategy towards building that agent that OpenAI has already been touting for a while now. So that wraps it up for this video. What do you think? I'd love to know what you think about ChatGPT tasks, how you plan on using it. Is it overhyped? Let us know in the comments. Also please do like this video and subscribe to the channel. It really helps us and with that thanks for watching. Sam.
