ai like im 5: what really is artificial intelligence? (article 2)

anyways,

imagine this, 50 years from now you are sitting in a war bunker during the first robotic world war. you think to yourself, how did we get here! ai is going to shake our lives up but most people don’t really understand or get it. if you are apart of this group, i have some good and bad news for you

the good: ai is in its infant stages, we are still figuring out what it can do, and most modern ai’s might be incredibly stupid to what we have in 10 years

the bad: ai is in its infant stages, we are still figuring out what it can do, and most modern ai’s might be incredibly stupid to what we have in 10 years

• our attempt at creating human like intelligence in computers and machine
• these computers and machines can make predictions,
  interact with the environment, and do other stuff
• they learn from these interactions, getting better
  each time, adjusting their actions or predictions

a key idea to computers and machines is that: they do simple things very very very very fast.

• this is important idea because the combination of simpler things can lead to complex things!
• computers do not have a brain, they operate purely based on a bunch of little switches that turn off and on

one of the cool things about humans is our brains are complex and adaptable, but still made of little switches turning off and on called neurons.

• it is impossible to recreate the human brain, but instead it is possible to achieve tasks the human brain can do, like learning, pattern recognition, and problem-solving.

• even the smartest ai’s in the world today, started out very dumb.
• so how did they get that point???

learning

i am going to give a task to find my favorite number 1–10000

• if you get it correct you win a million dollars, you get unlimited guesses too!
• each time you guess, i am going to give you a hint, lower or higher
• as humans, you can probably recognize there are ways to solve this problem

1. starting at 1 and guessing every number until you reach the number
2. starting at 5000 in the middle and guess the number that is halfway
3. randomly picking numbers and hoping you get it right

notice

• there is an optimal and fast way to solve this problem and win this million dollars
• you guesses are going to get better and better if you use this optimal way
• if you don’t you have no idea if you are getting better and better

this is because our optimal way incorporates learning, we are learning from our previous predictions and using them to affect our current and future decisions!

• now let’s say i asked to redo it, but instead 1–1000000!
• would you really find get the million dollars without incorporating learning!

so helpfully understand the cycle of this problem

• make predictions — — — — → see if correct — — — — → make best correction possible
• you can think of this a progression from dumb to smart, where we are learning each time.

here is a graph to illustrate this progression from dumb to smart, when we incorporate learning we are becoming smarter and smarter each time.

this is important idea, no matter what task an ai does and how complex it is, it has to start out dumb and requires this learning to improve

• simple ai’s are created to be excel at learning very simple stuff!
• complex ai’s are created to excel at learning more complex stuff!
• math, statistics, and other fields allow us to find ways to excel

“do one thing and do it well”

• the vast majority of ai’s are focused on 1 task or minimal tasks.
• we are still in the infancy so this approach is much more practical with our current knowledge
• that is okay because we can take these simple ones and build complex ones with them!

these ai’s are called narrow ai and here are some below!

1. computer vision

• ai attempt at visualizing the world using pictures, videos, sensors, etc
• there are a lot of types of computer vision such as facial recognition, object detection, and image classification

here drawing about object detection!

2. nature language procession

• ai attempt at understanding and mimicking human language and interaction
• this can involve things like chatting, talking, hearing, etc
• chat gpt, alexa, siri

here is a illustration of me and my alexa

3. robotics and ai

• robotics and ai have integrated to attempt to mimic human decision making systems
• can complete tasks like running, making coffee, and taking your job
• you may have heard the term automation with many jobs and this is what it refers too!

4. predictive analysis

• ai attempt at recognizing patterns and making predictions about the future
• allows us to make informed decisions
• think of this as predicting the price of a car in 10 years, or classifying a email into spam or not spam.

so the majority of ai’s are focused on 1 or minimal tasks what about that other!

these are called artificial general intelligence (agi) and here is an image to describe them

agi

• when people say they are scared of ai, this is what they really are scared about
• these ai’s are not limited to one task, but instead possess general knowledge in different fields
• they can learn new skills, interact with new environments, and not explicitly told to behave a certain way
• i am going to go more in depth of this in a future article!

hopefully things are starting to click and you understand a good amount about ai. i am going to be covering the following topics in my next article.

• data science
• machine learning
• deep learning

thank you for reading this, hope you have an awesome day, here is a picture of a quokka!

godspeed