Lamin Rubas, an AI upskilling specialist at the Applied AI Initiative, introduces the concept of generative AI in the first video of 'How to Master AI Tools at Work: The Generative AI video-series'. He explains what generative AI is, how it works by learning from vast amounts of data to generate new, similar data, and its practical uses in various fields. The video covers the generative capabilities of AI in inspiring ideas, summarizing market research, transforming text formats, assisting in coding, and more. Rubas highlights that generative AI is multimodal, capable of mapping many input formats to different outputs, and delves into the fundamental principle of how generative AI makes predictions based on probabilities. He concludes by inviting viewers to engage with the content for more AI-related videos. Explore more about Lamin Rubas's profile here: https://www.linkedin.com/in/lamin-rubas-520081143 Let's build a community of learners and enthusiasts together. Visit our website: Applied AI Institute Don't forget to like, share, and subscribe for more insightful content!
Hello AI learners, I'm Lamin Rubas. I'm an AI upskilling specialist at appliedAI. Welcome to our Genitive AI video series, How to Master AI Tools at Work. If you're curious about Genitive AI, if you're interested in how it works, and want a practical introduction into some very useful everyday applications, then obviously you're in the right place.
So, what exactly is Genitive AI? Let's find out. In very simple terms, it's a type of artificial intelligence that can generate new data, such as text, images, or videos. And it does that often in response to prompted instructions or questions that you give it. And how does it do that?
Well, put simply, a generative AI model, a model is basically a specific version of a generative AI, learns patterns in the vast amount of data that it is trained on. And after this training process, the model can generate data that has similar characteristics. So in case of a generative language model, it might be trained on a lot of text on very different and a large amount of internet web pages.
And there it can pick up on such things as grammar, some facts about the world, or also how specific people who might write text on those websites. This, it can then reproduce in new ways to generate text that seems useful to us.
Now, why would I care about this? Why would I care about generative AI? Well, we could make use of these generative capabilities to inspire fresh ideas or designs. to maybe summarize results from a market research, to easily transform text formats, or also assist in writing code, or help us with that super formal email that takes you way too long to figure out the right words to use for.
If applied smartly, Different generative AI applications can really complement you in your work. So you can focus more on the tasks that really require your attention. And in this video series, you will get practical recommendations and guides on how to do so.
One thing to note is that generative AI is multimodal. That means it can represent mappings from many different input formats. Two different output formats. That means we can input maybe a text based prompt, so a sentence or a question, to get a text based answer as an output. That's a very simple case.
Maybe you've already come across it. But we can also give a text based prompt to generate an image. Or use a few images, like this. as input to generate a short video sequence. So, there are really many different types of outputs that can be generated from many different types of inputs, sometimes even within the same application.
Now let's look a little bit under the hood and see on what principle generative AI, this kind of generation that happens, is based on. Like most AI models, the output is based on probabilities that the model learns somehow. Fundamentally, The only question the generative AI model answers is given a specific input, so something that goes into the model, what is a likely output?
And this works wonderfully for sequences of things. And these things could be words, they could be pixels, or musical notes. So the most probable continuation for this word sequence, the most beautiful city in Germany is, could be München, but it could also be not located in Bavaria. Now imagine we cut off the sentence somewhere else, like here, or here.
So, really at every single step in this word sequence, at every single word that comes in this sentence, the model really answers the question, what is the most probable continuation given what came before it? So we look at the sentence or the word parts that came before it, And then try to predict what is the next word.
Do that at every step. Now you might rightfully ask, how does this model know the probabilities of what should come next? Remember, a generative AI model learns patterns in its training data based on complex statistical and mathematical methods. So it really looks at all of the data that was fed into it and kind of extracts.
What are typical patterns? What are typical structures that are represented in this data? And if the data is distributed in a way that includes, for example, many blog articles from people from Berlin, then you might already guess that the second continuation might be The one that the model rather predicts.
But if it's a model that has training data that leans more into text that favour Munich, then the first continuation might be chosen. And if the training data is very diverse, contains very academic texts, maybe many different argumentative texts, then it Might look something like this. Does that maybe look familiar to you?
Core principle behind how generative AI works is relatively simple. But its technical intricacies and its way how it chooses different outputs, those are really complex and sometimes difficult to look through. But it is safe to say that generative AI is a very powerful technology that can inspire creativity and innovation and really can help us with these tedious tasks that we have throughout the day.
From creating art, to music, to assisting in content creation. The possibilities are really vast. So That wraps up our first quick exploration of generative AI. If you found this video interesting, then give it a thumbs up and share it with your friends. And of course, if you want more AI content, let us know in the comments below and tell us what you want to hear.
So, thanks for joining the appliedAI Institute and we'll see you in the next one.