Recently, I’ve been working on my experience and knowledge surrounding the use of artificial intelligence, and more specifically, artificial intelligence models. There are two types of artificial intelligence models that I learned about : discriminative and generative. Discriminative models focus on differentiating, predicting, and labeling data. By learning boundaries, they divide (differentiate) the data space into classes. For example, if you ask a discriminative AI model if an email is considered spam, the model would be able to tell you if it is based on the relationship between the input’s variables and the target variable.
Generative AI models are different in that they can actually generate new data samples based on the data it’s trained on. While a discriminative model can only label an email as spam, you could ask a generative AI model to recreate an example of a spam email. The model can do this because it understands how one data point relates to other data they are trained on. You can visualize this in a sort of spiderweb shape, where there are million vectors interconnecting each other. At the end of each vector, there is a single data point, word, or idea. One idea can be connected to many others. This is how generative AI models work, by using this data network to create (generate) new sets of data for you.
One thing I’ve learned about when exploring generative AI models is prompt engineering basics. With the release of Chat-GPT and many other generative AI models, people have been able to accomplish and create great things with simple prompts. I used a prompt engineering guide I found to teach me the best ways to create prompts for AI models, where they can understand it easily and give me exactly what I am asking for. I’ve learned a lot of different techniques for it that I didn’t know existed, and it’s helped me a lot. You can find the Prompt Engineering Guide here:
https://www.promptingguide.ai/
The second thing that I have discovered is Partyrock, which is an Amazon Web Services website, where you can create your own artificial intelligence models really easily. You don’t need any deep-level knowledge of data science or anything, you can just enter in a prompt of what kind of AI application you’d like to create, and the website will do it all for you. Then, you can edit your AI model and play around with it before publishing it. I created a fun Vacation Planner, where you enter a location, time frame, and budget, and then the AI model will give you a full budgeted and scheduled itinerary for your vacation. I also added a chatbot to ask questions about your vacation, a list of fun facts about the destinations, and a section that generates images of the destination. You can find my Vacation Planner here: