AI Applications and Use cases
AI Applications and Use cases
With all the talk about AI, lets try to understand what does AI do?
We can generally classify AI applications and tasks into 4 types
- Prediction
- Classification
- Automation
- Generative
Prediction AI
When an AI is trained to predict a outcome based on previous occurences, we consider it as prediction AI. Good example of this is many Stock Prediction AIs. Infact my final year project also made use of Prediction AI models to help analyze and predict stock price. We also come across a different type of prediction AI in our day to day life, which we just use the term "algorithm". I'm refering to Social media algorithms. You might have encountered that you watch a certain type of video on youtube or a type of reel on instagram, and after sometime you notice more similar kinds of posts/reels/videos popping up on your feed. This is due to the AI trying to understand what you might like and suggesting stuff based on previous watched content. As time goes the algorithm understands what exactly are your preferences, what you are most likely to click on, what you are more likely to spend your time on, what kind of channels or accounts you like, the type of content et cetra. This is due to the AI making a tailored feed for you.
Classification AI
When an AI is trained on a set of data which reperesents different classes, and then the AI can distinguish a given data and classify it accordingly, it is seen as a Classification AI. To understand this in a simple way let me put it like this. The AI model is given a bunch of pictures of cats and dogs which have corresponding labels. The AI uses the labels and analyses the content of the pictues to make a connection between the content and the label. Then a set of data of unlabeled pictures is given to the AI model and we ask it to distinguish. It understands the characteristics and differentiates it. This type of AI can be seen being researched and trained in many fields, one of which is the medical field where the AI can be fed a series of symptoms to shortlist the possible diseases. This kind of AI can make a human's Job easier as it can save them valuable time, and in the field of medical science, this saved time can help save many lives.
Automation AI
Automation AI uses the above 2 categories to automate a human process. A good example can be seen again in the algorithms of social media and other web pages. Let us take the example of the internet websites and Ads. There might be situations where you look up a laptop on a search engine, and check a site or 2 but then move on. Minutes later you will find advertisements of attractive deals of exactly what you want. This is done through automation AI where the system is able to understand user clicks and patters, classifies the data, and then uses that to predict what user may need.
Generative AI
Generative AI is one of the newer innovations in the space of Artificial Intelligence and Machine Learning and it is quickly growing and becoming more and more advanced. This kind of AI uses a lot of previous data to understand data itself, whether it be text, image, speech, etc. Generative AI uses a series of different machine learning and deep learning algorithms to then generate some sort of media, whether it be text, image or speech
The generally seen generative AIs are Text generation, image generation, Audio generation. These can be used together to generate Animations or Videos and more.
A good example and probably the most well known example of a generative AI is ChatGPT. This uses something called as a Large Language Model to train itself the gpt model. Similarly an Image Generation AI like DALL-E, Stable Diffusion or MidJourney uses a large set of images from the internet on which the models are trained.
Usually the method on which the generative AIs work is the user gives in a prompt describing something or asking a question and the models use this vast amount of data to process and give a outcome which is human like.
Comments
Post a Comment