Test AutoGPT for FREE & Other Cool Open Source Projects

This is pretty neat. So it's a detailed report generated by AutoGPT, but for absolutely free. In this video we will be covering some of the exciting new open source projects. For example, how to do market research with Auto GPT or create cool memes with AI. Or create a personal assistant with AR glasses. Or give memory to ChatGPT so it remembers all your previous conversations. At the end of the video, I will show you how to use R GPT for absolutely free. So let's get started. If you're looking for how to integrate RGPT in your workflow, here is an example. This person got an AI agent to do market research and it wrote reports on its findings. He pretended to have a fake shoe company and got AutoGPT to do a simple market research for him. So, so the objective was do market research for waterproof shoes and get the top five competitors and give me a report on their pros and cons. The agents were able to do research automatically and at the end it generated a pretty awesome report based on the actual brand. So for example, you see Columbia, Salman Barrel Keen, the North Face, right? And every competitor has pros and cons. Now the great thing is, according to this person, this whole process took only 8 minutes and 10 cents. If you want to take your meme game to the next level, there's an awesome tool for that. Now introducing Meme Cam. All you have to do is just upload an image and it will generate a meme out of it. Okay, so I uploaded this image and it actually is pretty awesome. So finally a room that matches the size of my ebook. This is pretty neat. The first project we're going to look at is called Memory GPT. It's basically chatgpt, but with long term memory. Here is how it works. First, you have a conversation with the chatbot. It's called Brainy. Then we'll simply click Remember Chat and restart. And when you restart it the next time, it will remember all the conversation that you already had. Now remember, this is absolutely free for the time being. So here is an example conversation that I had. I told the chatbot that I'm starting a YouTube channel focused on AI and other cool technologies. Okay, now if I want the bot to remember my conversations, I'm going to go click Remember Chat and restart. It will save the conversation. All right, let's see if it actually remembers a live conversation or not. Asked me for you. I told them I'm good. And then the traditional as an AI agent, I don't have feelings and everything, right? Okay, so let's say do you remember. Remember what we talked about last time? Okay. Let's see what the response is. All right. Okay. So to remember, as you mentioned, that you are starting a YouTube channel. Focus on AI and other cool technologies, mostly tutorials. It's a pretty cool tool, especially like if you want the chatbot to remember your conversations long term and you don't really want to prompt it again and again. Just keep in mind it's free, but make sure that you're comfortable with their privacy policy. I didn't see anything. So play with it at your own risks. I personally really love this project. So if you are not familiar with archive.org it's a free distribution service in open source archive for the latest research in the field of physics, mathematics, computer science and machine learning, etc. Researchers post their latest papers free of cost on this platform. But now with this new tool, they turned it into a podcast. So rather than reading a paper, now you can actually listen to it as a summary. And so let me just play one paper. This. Basically, it's going to be something like this.

Even when the explanation and user agree there may be hidden flaws in the model. The paper proposes a new method called Explanation Guided Exposure Minimization to mitigate these hidden flaws by pruning model variations that do not receive positive explanation feedback.

So if you want to be familiar with the latest research, but you don't really want to read all the paper, here is a way in which you can simply listen to it as a podcast. And it's really awesome because it summarizes each paper and its findings. Here is a really interesting project. So this person created a personal assistant that is connected to augmented reality smart glasses. Now, the way it works is it uses computer vision to recognize faces of your friends and then based on your chat history. So for example, it has access to your text messages and everything. It brings up relevant topics and details that you can discuss with them and it simply displays everything through your smart glasses. This is pretty cool and a little uncanny. A few days ago I covered this paper on AI agents behaving like humans by Stanford and Google Research. Now, this person put that into practice and actually produced this demo of how, let's say NPCs will interact in future with humans. So, so I'm going to play this a little bit. The first is the human voice, and now the AI is generating the response. He can be found in his smithy.

Near the old town square.

If you're looking for Drake, then you should talk to the Local TAVERN KEEPER Interactive NPCs are already coming into games. This would change the gaming industry forever. If you want to experiment with RGPT but don't have access to GPT4 yet, there's a pretty neat website that you can check out. It's called Agent GPT. Keep in mind, for the time being it's free to use, but the team is actually burning through cash. So I would encourage everybody, if you can, please support them. All right, so let's run a quick test. So you need to provide the name of your agent. In this case I'm providing researched GPT and the goal. So video and in this case I'm saying video ideas for YouTube channel on the current advancements in generative AIs. Right? Then you need to simply click Deploy Agent and it will start deploying, right? So first here's the task or the code that we give it and it's thinking. Then here's a list of tasks that it came up with. So analyze current trends in developments in generative AI through web scraping and data analysis. Then the second task is generate multiple possible video topics based on the finding and analysis of the first task. And then another task, the task three. You can actually see the task here. So utilizing utilize natural language generation to create detailed outlines for outlines and script for the chosen video topics. Okay, so it simply goes picks one task and then start executing through it, right? So for example, it's executing the first task, then it's thinking and then started adding the second task to the list and started executing that. Right? And then the third task. So basically the auto GPT picks a task, then spins up an agent to complete that task. Right? So that. That's how it works. Now keep in mind that this is a demo, so it's not going to run forever and it will stop executing. Right? But the great thing is, if you have access to OpenAI's TPT4, you can include your own API keys here to increase the limit. Right? So like for example, here you sound sorry, we're sorry because this is a demo. We cannot have our agents running for too long. Note, if you desire longer runs, please provide your own API key in the settings. So shutting down. Okay, now the awesome thing is you can actually save the report that it generated so far. Come and click on save. Right? All right, so here is all the steps that it took, right? And it actually also included its findings as well. So for example, when it was executing this third task, it came up with a list of topics to cover. So gans for image and painting stylegan2 for art generation NLP for summarization auto encoders for image compression Reinforcement learning for game development. These are pretty good topic actually. I might pick some of these and start making videos on those. So wise cloning with deep learning. I also have already have a video on this topic. GPT3 for creative writing neural style transfer object detection with YOLO. This is pretty awesome. Pix to pix for image to image translation. I actually have a video on this topic. Okay, these are some awesome. Yeah, so it's it's pretty useful tool. All right, so this was a quick video. Just wanted to cover some amazing open source tools. Links are going to be in the description of the video. Hope you enjoyed this. Thanks for watching. See you in the next one.