TKS AI Module Wrap Up

Hi everyone, I just finished the TKS AI module (stay tuned for a couple more AI module-related content to fully wrap up the module). I know that I haven’t done updates on my blog (just for modules though, everything else is still on schedule). Just go to my LinkedIn here to have access to my profile and past module posts, including posts on the new module I am starting.

Now, back to this AI module. I just spent most of this week doing the biggest deep-dive I’ve ever done. This was the most interesting thing ever. I learned about AI, from what it is made up of, different models and learning types, different uses and applications, and different companies that are utilizing AI in the most interesting ways. I’m going to cover what makes up AI and what I’m most excited to see in AI advancements.

The first thing I wanted to cover in this blog post is the overall difference between AI, ML (Machine Learning), Neural Networks, and Deep Learning. All of these are the basis for another, but it’s more complicated than that (not actually complicated).

Let’s begin with AI, the executer, or the face, of what we see when we use these tools. AI executes the prompts we give it to create all of these amazing things that we can apply to whatever we want. AI is the generator for content that we ask it to make. But AI isn’t just AI, there is something within AI that does a lot of the learning and searching through data sets to match your request.

The layer underneath AI that does the learning and analysis of data sets is ML. ML stand for Machine Learning. Basically what that means is this is where all of the learning and understanding of data, information, or feedback takes place. ML takes all the resources from what it’s learned in that area and can take it to the AI to generate content or execute tasks based that info. But ML isn’t just this one place where everything takes place in one magic blob. ML is then made up of neural networks.

Neural networks in an AI brain are largely based off the neural networks in our brain, where everything is connected and all that information goes through our mind to be sorted out or to generate new ideas (like AI does). Neural networks have three layers:

Input Layer: The layer where all the information goes in from. That can include data sets, resources, or feedback.

Hidden Layer: This is where the neural networks can sort through the information and create these connections to learn and process it.

Output Layer: This is where all the information connected, learned, and sorted in the hidden layer comes out for the AI to use it.

All these layers make up the Machine Learning process. But there are 3 different learning models: Supervised, Unsupervised, and Reinforcement Learning. You can read about that on my LinkedIn.

Neural networks can be made deeper into deep learning. Deep learning, essentially, is a more complex makeup of neural networks. You can classify deep learning as a connection of neural networks witn 4 or more layers. The extra layers would be added in the hidden layers. You might add two or three sets of hidden layers, and that will classify it as deep learning.

To recap, you have either deep learning or neural networks. Whichever set up you choose will be the process for ML, which will then take what it’s learned and send it to the AI to execute tasks based on the ML.

AI has advanced so much in the past years, from the way we can teach it to all its applications in different fields. One company that is working with AI that I find super interesting is Google.

Google has been incorporating AI into all of its existing software and tools. It’s revamping your Google search with personalized recommendations based on what you’ve searched and making the entire experience more interactive. But what’s really interesting is seeing everything that Google DeepMind in working on. I love that they are doing constant experimentation to create these robots and AI brains that are actually changing the world and helping to solve problems.

For example, Google DeepMind participated in CASP, which is a opportunity for groups to test protein structure prediction menthols, which has been a big problem in biology for a while now. But Google DeepMind was able to create an AI tool that was able to predict these protein structures much more effectively compared to other groups in the past. And that’s just one of many examples that are so impactful.

That’s my recap of the TKS AI Module. I am currently working on the motivation and time management module to hopefully get some tips on how to manage all of my tasks and commitments. Might as well do it now so I have these tools for the rest of the year. Check out my LinkedIn to see updates on that module!