Ep05: Build an AI Desk Companion with Quarky Intellio | ChatGPT Integration | PictoBlox

Duration: 6:00 912 transcript words EN summary

Video summary

In this episode of the Quarky Intellio series, the focus shifts from basic hardware interactions to leveraging advanced artificial intelligence by integrating large language models like ChatGPT into the robot. The video demonstrates how connecting Intellio with PictoBlox allows the device to move beyond simple command execution, enabling it to think, converse, and answer questions naturally. By utilizing extensions for speech recognition and AI interaction within the PictoBlox environment, users can program their robot to listen to voice commands, process them through an external server, and generate human-like responses that are then spoken aloud via its speakers. The technical workflow behind this integration involves a seamless pipeline where Intellio records audio using its microphone, which is then converted into text via speech-to-text technology. This text is sent to a server acting as a bridge to the ChatGPT API, where the model processes the query and generates a textual response. The system then converts this text back into audio using text-to-speech capabilities before playing it through the robot's speaker. Through this process, the video showcases various applications such as answering general knowledge questions like the capital of India or solving math problems, while also highlighting that the AI might occasionally need a second attempt to understand specific queries correctly. Beyond simple Q&A, the tutorial explores more dynamic features where the AI can generate jokes based on puns or function as an interactive quiz master. In the quiz scenario, the robot asks general knowledge questions and validates the user's answers in real-time, providing immediate feedback on whether the response is correct or incorrect. These examples illustrate how modern AI assistants are built using fundamental concepts like APIs, servers, and processing pipelines, effectively turning a standard educational robot into a personalized desk companion capable of engaging in meaningful conversations and learning activities. The video concludes by encouraging viewers to experiment with creating their own personalized AI desk companions while emphasizing the importance of responsible AI usage through controlled prompts and safe API practices. It serves as an introduction to real-world AI concepts, bridging the gap between theoretical models and practical application in a tangible device. As a preview for future content, the host hints at upcoming lessons on using Intellio in machine learning, inviting the audience to like, subscribe, and stay tuned for further exploration into the evolving capabilities of intelligent robotics.

Read the full video transcript

Hey, hey, buddy. Can I tell you a fun fact? >> Of course. Howdy. >> One day on Venus is around 8 months on Earth. >> Okay guys, enough talking. Now, let me talk. Hey. Hey, everyone. This is AI and welcome back to our quirky Intelio series. So far, we have seen Intel streaming using its camera and interact with microphone and speaker. Now in this video we are exploring Intellio's intelligence using large language model like charged or Gemini. Are you excited? Let's get started. I'm sure you know that charge GPT is a generative AI model that understands natural language and generates humanlike responses. So by connecting Intellio with chat GPT your robot can think, [music] answer and converse and not just follow commands. Let's see how this will work. So as a first step we will switch on Intellio and connect to Pictoblocks. Now you already know how to connect Intel with Pictoblocks with first method which is via Wi-Fi router. And if you don't know how to connect, you can obviously watch this video for more information. The other three methods we will be understanding in one of the upcoming videos. So hold tight. I have now connected Intellio with Pictoblocks via Wi-Fi router method. So let's start coding now. So as a step one in picto blogs first we'll get this two extensions which we are using today speech recognition and chat GPD. Once that is added let's get a event when green flag clicked. After that we will go to speaker blog [music] and set the language to English. Now we will go to chat GPT extension and get this block where we will set the maximum length of words to 15. You can obviously set it to whatever you need but currently we're setting it to 15. Now we will go to speech recognition block [music] and from here we will get this recognize the speech for 5 seconds. I'm going to keep it to 5 seconds. So it will take my response and I'm going to also make Toby say whatever I have said. For example, what is the capital of India? After that I will send this response to chat GPT to ask a question. So we will take this block from chat GPT ask AI and put this speech recognition result block over here. Then I'm going to go to speaker block and make Intelio speak whatever the response chat GPT has given. So this block will be used get AI [music] response and the same thing I will make Toby say as well. Wow that was simple right now let's [music] check if it works. What is the capital of India? the capital of Indonesia in heaven. >> Wow, that's great. Right now, similarly, you can create a story board or a study buddy and you can ask any of your questions which you're curious about to chat GPT like this. This is your own personal assistant. Now, how this works in the back end, let me give you a little bit detail about it. First, Intellio records audio using its microphone. [music] Now, that audio needs to be converted into text via speech to text. This text is sent to server via API request. The server acts like a bridge between chat GPD and Intellio. [music] Chat GPD processes this voice prompt and generates a response in form of a text. Now that text is converted into audio via text to speech and that audio is played by the speaker of Intel. So the entire [music] pipeline looks something like this. This makes Intellio an AI desk buddy that can answer any question. Let's try a math question. What is the square root of 144? >> I'm sorry, but I cannot provide a response to know. >> There is a possibility it may not understand first time. Let's try it again. What is the square root of 144? >> The square root of 144 is equal to 12. >> That's correct. Or it can also dynamically generate a joke. Let's try that. Can you tell me a joke? >> Why did the scarecrow win an award? >> Because he was outstanding in his field. That was a good one. Right now, we can also try one more project. Check out this quiz master which I have created where it will ask me a question of general knowledge and once I give an answer, it will check if it's correct or not. So let's try. >> What is the largest planet in our solar system and what is its name? >> Jupiter. >> Correct. >> Wow. Now let's see your next question. >> What is the capital of France? >> Paris. >> Correct. What is the name of the closest star to Earth? >> Let's try a [music] wrong answer. Pluto. >> Wrong. Wow, that's interesting, isn't it? You are now working with real world AI concepts [music] like APIs, servers, AI models, pipelines. These are the same building blocks used in modern AI assistant. Now, I have a task for you. Try to create your own personalized desk companion like this by using chat GPT. Remember, we use control prompts, filtered responses, and safe APIs for responsible AI use. In the next video, we will learn how to use Intel in machine learning. So, make sure you like, subscribe, and hit the bell icon. [music] Until then, happy learning.