AI, the know

What Exactly is AI? 

AI, or Artificial Intelligence, is in essence giving machines a brain similar to ours, allowing them to think and make decisions. So, what does that really mean? Imagine we teach computers and programs to think and solve problems just like we do. Let’s use a simple example: your smartphone is filled with photos, including many of the same people like your family and friends. An app with AI can recognize these faces, similar to how you can, and know who they are—your mom, dad, sibling, or friend. This app knows this because it has learned to identify each person’s face. A great example of this in action is Google Photos, which is even more clever and capable than this simple explanation suggests.

AI can learn?

AI can indeed learn and there are two main aspects; machine learning and deep learning. Let’s start with machine learning.
When we want a computer to do something, we use programs to give instructions to the computer. You might have heard some programming languages before; Python, Java, Go, JavaScript, etc. These are the languages that we use to give instructions to the computer. But, what if we didn’t have to type every single step of the instructions out to the computer? What if we wanted to give some instructions to the computer, some data for it to reference, and told the computer to, “figure the rest out.” That’s what machine learning is. Instead of telling the computer every single step of the instructions, machine learning enables computers to automatically learn without being explicitly programmed for each specific task; this ability comes from statistical techniques and complex algorithms.

Deep learning is like the neural network of the human brain. There are many layers to deep learning, each layer utilizing algorithms that intertwine in complex ways which allow for abstraction on many levels of the layer. That’s a lot to wrap your head around, so here’s an easier way to look at it. Deep learning uses many layers of learning, similar to stacking up lots of layers of building blocks. Each layer helps the computer understand better. Similar to how you build with blocks, the more layers you have, the cooler structures you can build.

AI is created to solve specific problems and it uses cutting edge technology to solve those specific problems. Through machine learning and deep learning together, AI can discover patterns using vast amounts of data, at a scale and speed that humans are unable to match. AI is good at analyzing data and evaluating complex scenarios and conditions to choose the best course of action, all based on data. An example of this is perception. Perception means to become aware of something through senses. Now if you’re thinking to yourself, “AI doesn’t have senses!” No, but we do and we can train AI to perceive the world how we perceive it.
We do the same thing when we teach our kids. When they ask what an elephant is, we describe it to them and nowadays show them a picture. In a similar way, when AI needs to be taught what an elephant is, human based input confirms what is and is not an elephant – that is a human sees an image of an elephant and clicks / touches yes or no and that response is fed to the AI. There are other ways to train AI, but you get the point

But how does the computer learn?

There are three learning models; supervised, unsupervised and reinforcement. Supervised learning means the AI model is trained on a labeled dataset. That essentially means that each example or “piece” of data is labeled with the correct output – almost like a sticky note with the correct answer written on it and that sticky note is placed on the data item.

Unsupervised learning involves training the model on data without explicit “sticky notes” but instead enabling the model to identify patterns in the data itself.

Lastly, we have reinforcement. This essentially means that the AI has received feedback on the output in the form of rewards or penalties; rewarded for correct outputs and punished (naughty computer!) for bad outputs.

Data as a teacher

Just as humans learn, AI must learn too. It is the foundation of how accurate and predictable the AI will become. AI requires incredibly vast data sets to build the “intelligence.” Because let’s be honest, there wouldn’t be much practical use for AI if it didn’t have the intelligence piece, although it could be quite entertaining. Imagine if you asked chatGPT a question and it replied with, “Hell if I know what coffee is made out of.” I digress. AI uses data sets to learn, understand, and (and hopefully) correctly predict the outcome of the task it was given. A trivial example could be the following; you want to build an AI that correctly identifies pictures and gifs of cats. By feeding the AI millions and millions of cat pictures (supervised learning) you can train the AI to accurately identify cats within an image or gif. That’s one of the powerful aspects of AI, it is able to process data more efficiently and MUCH faster than humans can.

How do we use AI?

AI has much more powerful use cases than identifying pictures of cats.It is leveraged in many different sectors; health, finance, security, etc. Let’s start with the health sector. For example, AI has helped diagnose diseases and cancers, utilized in fraud detection systems for finance, and enhanced security using image and video recognition tools. The extent to how we can utilize AI extends to how many problems we have as species, which are A LOT of problems. Of course AI cannot solve certain problems like world hunger or climate change, however IT CAN help us to understand how we can mitigate or reduce those problems; developing more efficient grains using bioengineering, or finding rare ionic combinations that are stable and more suitable for battery technology. The key here is that AI is powerful at assisting with specific problems.

In short

We are currently in the gold rush of our time, but the gold is AI and the power behind it. AI has already disrupted the status quo of many industries and many would argue that the age of AI has only just begun. With the power of tools like chatGPT from OpenAi, Google Bard, and Microsoft Copilot, many tasks have been streamlined, from content generation to writing to coding and everything between. If you have not tried using AI yet, I highly encourage you to try ChatGPT – the entry barrier is near non-existent as you only have to sign up (it’s totally free though there is a paid version but it’s not necessary nor will we go into depth on the differences between the free and paid versions).

Personally, with the power of AI at consumers’ fingertips, I believe we are at a paradigm shift of society and I am excited to see the development. Let me know what you think about AI, how you’ve used it thus far, tools you’ve made, custom GPTs you’ve developed, or anything else you find notable with AI.

01101000 01110101 01100111 01100010 01111001 01110100 01100101 01110011