What Are The 4 Types Of Ai?

Did you decide to give it a go in trying to understand what AI is, how it’s growing and also understanding its differences?

Currently – 2021, can you classify the types of AI? Well, this might sound so difficult, so if you are not a tech nerd but whatever your level in AI, it’s a good idea to understand how technology is evolving and stay updated.

In 2021, we can classify AI into 4 different types. Well, the types are more of the Maslov’s hierarchy of needs whereby the simplest level only needs basic functioning and the most advanced level is where things like all knowing, all seeing and also self-aware consciousness are needed.

What are the four types of AI?

The four types of AI that you need to know of in 2021 are:

1. Reactive Machines

2. Limited Memory

3. Theory of Mind

4. Self Aware

All you need to know now is that the AI level has completed the first type and is also improving and perfecting in the second type. Currently, the third and the fourth types of mentioned above only exist in theory form.

They are expected to be the next stage of AI. What do you think? Let’s find out!

Reactive Machines

Ever heard of reactive machines before? If not, they are used to execute basic operations. This is also considered as the simplest level of AI. Reactive machines react to some input with also some output. Here, no learning occurs. When designing any AI system, this is usually the first stage.

For example, a machine learning that takes a human face input and then outputs a box around the face to determine if it’s a face, is a simple reactive machine. This model is also known to store no outputs and also doesn’t perform any learning.

Static machine learning models are also reactive machines. Their architecture is also known to be the simplest and they are widely available in GitHub repos across the web. You can choose to download these models, trade or even pass around and load into a developer’s toolkit with much ease.

Limited Memory

Here, AI’s ability to store the past data and also predictions to use them in making better predictions is tested. With Limited Memory, machine learning architecture is said to be a bit complex. Again, all machine learning models need limited memory to be designed but also the model can be deployed as a reactive machine type that we earlier discussed.

You should also know that there are three major types of machine learning models that can be used to achieve Limited Memory type. They include:

- Reinforcement learning

- Long Short Term Memory (LSTMs)

- Evolutionary Generative Adversarial Networks (E-GAN)

Reinforcement learning

Well, these models are known to make better predictions based on several cycles of trial and error. Have you ever played any computer game such as Chess, Go, and even DOTA2? It’s this model that is used to teach computers how to play games such as the above mentioned.

Long Short Term Memory (LSTMs)

Experts and researchers intuited that the previous data can also be used when predicting the next items that follow a given sequence, more so in language hence the birth of this model.

Evolutionary Generative Adversarial Networks (E-GAN)

The E-GAN boasts of a memory that enables it to evolve at every evolution. This model produces something similar to growth. Growing things don’t usually take the same path every time, hence leads to the path being modified due to statistics is a math of chance and not math of exactness. When modifying, the model might come up with a path that has less resistance.

How a Limited Memory AI works?

There are only two ways that Limited Memory AI works:

- A team consistently trains a model on a new data

- The AI environment is built in that the models automatically train and renew upon model usage and also behavior

The most common is the Active Learning in the ML lifecycle. The ML Active Learning Cycle has six steps which are:

1. Training Data. There must be data used for training.

2. Build ML Model. The model is designed.

3. Model Predictions. The model makes predictions.

4. Feedback. Model finds its feedback on its prediction from human or even environment stimuli

5. Feedback becomes data. Feedback is taken back to a data repository.

6. Repeat Step 1. Continue repeating this cycle.

Theory of Mind

As earlier discussed in the introduction part, we haven’t reached the Theory of Mind in Artificial Intelligence types. Currently, we are seeing their beginnings such as self-driving cars. All you need to know is that at this given stage, AI starts to interact with the thoughts and also emotions of humans.

As you read this article, machine learning only does a lot for a person directed to achieve a given task. Models available today have a one-way relationship with AI. This is just like how Alexa and Siri respect all human commands.

For instance, if you angrily yell at Google Maps to change to another direction, it lacks emotional support and will only say that “this is the fastest direction. Who may I call and inform you will be late?” Again, Google Maps will also continue returning the previous traffic reports and even ETAs since it doesn’t know you’re already angry.

Self – Aware

This is the last type of AI that we mentioned earlier, right? Well, this one too doesn’t exist and hopefully in the future, AI may achieve nirvana. This is when it ends up being self-aware. This kind of AI is only present in stories and as you know, stories do trigger hope and also fear into audiences.

Try to imagine a self-aware intelligence that surpasses humans and also has an independent intelligence. This seems good but also bad, people will have to discuss the terms of such a system.

Are there other AI types apart from the four types?

Yes, there are other types and are known as crowd observers. They include:

- Artificial Super Intelligence

- Artificial General Intelligence

- Artificial Intelligence

Last Words

In whichever way that you may categorize AI, you need to keep in mind that AI is a software tool that has begun and is here to stay and also in the future. This may also lead to several thoughts – both good and bad. You only need to hope for the best and embrace technology. Goodluck!