AI | How ChatGPT Works? Three Videos to Watch.

Hi All,

AI continues to make big headlines. Before we get into the nitty-gritty details, I wanted to share some informative and educational videos that I found helpful in understanding the framework behind the 'generative AI' and 'machine learning' that chatGPT is based on.

Okay, you might ask what generative AI is. So, I asked chatGPT for an explanation for a 5-year-old kid or a senior person who doesn't know anything about AI. Below is what I got.

Q: What is generative AI?

A: Generative AI is like a smart robot that can make new things all by itself, like drawing pictures or making music. It looks at lots of examples of pictures or music, and then it makes its own new ones that look or sound similar to the examples. It's kind of like when you practice drawing or playing an instrument, and then you start making your own pictures or songs.

People can use generative AI to make all kinds of things, like cartoons, movies, or even stories. It's a very clever tool that can help us make all sorts of new things that we might not have been able to make before.

Q: Are there other types of AI besides generative AI?

A: One type of AI is called "rule-based AI", which uses a set of predefined rules to make decisions or take actions based on input data. This type of AI is often used in expert systems, which are computer programs designed to mimic the decision-making abilities of a human expert in a specific domain.

Another type of AI is called "machine learning", which involves training a computer algorithm to learn patterns and make predictions based on large sets of data. This type of AI is often used in applications such as image recognition, speech recognition, and natural language processing.

A subset of machine learning is "deep learning", which uses neural networks to model complex relationships between data points. Deep learning is often used in applications such as computer vision, where it can help identify objects in images or videos.

There are also other types of AI, such as "evolutionary algorithms", which use principles of natural selection to optimize solutions to complex problems, and "fuzzy logic", which allows computers to reason about imprecise or uncertain information.

I. Generative AI = Mimicking Human Brain Neurons?

One of the key things to understand about generative AI and computer programming is that it's an attempt by humans to emulate the connectivity of neurons in the human brain. In fact, as far back as 1905, Santiago Ramón y Cajal was already mapping out the neural pathways of rat brains and discovering that the connectivity of neurons varied depending on the location of the brain.

Some experts believe that 'intelligence' can be seen as a function of how neurons are connected, with different neural pathways and connections determining how different parts of the brain can see, think, smell, and move our bodies.

So, in computer programming, we are effectively trying to create 'artificial neurons.' The structure of neurons and their connections within the human brain have inspired the design of computer programming languages that utilize artificial neural networks. In the human brain, neurons communicate with each other through synaptic connections, where electrical signals are passed along from one neuron to the next. Similarly, artificial neural networks in computer programming involve interconnected nodes or "neurons" that communicate with each other through mathematical functions. These connections allow the neural network to process data inputs and produce outputs in a way that can simulate human thought processes. By modeling artificial neural networks after the brain's neural connections, computer programming has been able to develop sophisticated algorithms that can perform complex tasks, such as image recognition, natural language processing, and decision-making.

Thankfully, Fjodor van Veen from Asimov institute compiled a wonderful cheatsheet on Neural Network topologies.

II. ChatGPT Framework

The below diagram shows the AI framework for ChatGPT. I hope this introduction helps set the stage for the informative videos on ChatGPT's generative AI framework that you're about to watch!

III. Three Videos to Watch to Understand ChatGPT!

1. How ChatGPT is trained (14min): This is a short version.

2. The REAL potential of generative AI (20min): What is a large language model? How can it be used to enhance your business? In this conversation, Ali Rowghani, Managing Director of YC Continuity, talks with Raza Habib, CEO of Humanloop, about the cutting-edge AI powering innovations today—and what the future may hold. They discuss how large language models like Open AI's GPT-3 work, why fine-tuning is important for customizing models to specific use cases, and the challenges involved with building apps using these models. If you're curious about the ethical implications of AI, Raza shares his predictions about the impact of this quickly developing technology on the industry and the world at large.

3. GPT3 & Beyond (58min): Key concepts and open questions in a golden age for natural language understanding is explored in this webinar by Stanford Professor Christopher Potts. He discusses the significance and implications of recent Natural Language Understanding developments including GPT-3. He describes the fundamental building blocks of these systems and describes how we can reliably assess and understand them.

https://www.youtube.com/watch?v=-lnHHWRCDGk

BONUS - P.S. If you want to hear from the man himself - check out this interview. Greylock general partner Reid Hoffman interviews OpenAI CEO Sam Altman. The AI research and deployment company's primary mission is to develop and promote AI technology that benefits humanity. Founded in 2015, the company has most recently been noted for its generative transformer model GPT - 3, which uses deep learning to produce human-like text, and its image-creation platform DALL-E.

One more! What if your local pizza shop could use AI to predict which flavor would sell best each day of the week? Andrew Ng shares a vision for democratizing access to AI, empowering any business to make decisions that will increase their profit and productivity. Hear from the father of AI himself. Learn how we could build a richer society – all with just a few self-provided data points.