Today, we’re diving into the basics of Artificial Intelligence (AI) — what it is, what it can do, and how it has evolved over time. We’ll also explore practical uses of AI, with a focus on ChatGPT, a powerful tool for performing text-based tasks.

Prerequisites:

• Curiosity: Be open to learning new things.

What is AI?

1956
John McCarthy formally introduced AI in 1956, as systems designed to automate tasks requiring human intelligence. Early applications were rule-based, limited to tasks like language translation or data processing.

2000s

The Chatterbox Challenge was an annual online competition in the early 2000s designed to showcase and advance the development of conversational AI, specifically chatbots. It was one of the early venues for developers to test and demonstrate the capabilities of their chatbots in natural language understanding and interaction.

Key Features of the Chatterbox Challenge:

1. Competition Format:
   The challenge invited developers from around the world to submit their chatbots, which were then evaluated based on how well they could simulate human-like conversations. The chatbots were tested through text-based interactions, with judges posing a wide range of questions to assess the bots’ responses.
2. Evaluation Criteria:
   Chatbots were judged on their accuracy, humor, intelligence, and the ability to maintain coherent, natural conversations. The competition sought bots that could convincingly mimic human dialogue while avoiding awkward or nonsensical responses.
3. Participants:
   Both individual developers and teams participated, with many submissions being built using early AI frameworks like AIML (Artificial Intelligence Markup Language), ChaosML, Java and Php. These early chatbots could handle a variety of basic tasks, such as holding simple conversations, answering general knowledge questions, and performing basic commands (search engine, math, question and answers). Every rule was programmed by a creative software developer.
4. Significance:
   The Chatterbox Challenge was important for promoting the development of chatbots in the early 2000s, a time when conversational AI was still relatively primitive. It helped fuel interest in machine language, natural language processing (NLP) and conversational agents, which later evolved into more sophisticated systems like Siri, Alexa, and ChatGPT.
5. Notable Bots:
   Several well-known early bots, such as Jabberwacky and AliceBot competed in the Chatterbox Challenge. Jabberwacky was notable for its attempts at humor and casual conversation, and it later evolved into Cleverbot, which remains popular today. Onkwehonwehneha AI was developed by the author of this article (MoniGarr) and earned a Bronze Medal for Most Knowledgeable Bot for providing English to Kanien’keha (Mohawk Language) responses.

While the Chatterbox Challenge itself no longer runs, it laid the groundwork for future chatbot competitions and advancements in conversational AI. It was a space for AI enthusiasts to test the limits of early AI systems, pushing the boundaries of what chatbots could achieve at that time.

AI Today
On November 30, 2022, OpenAI launched ChatGPT, revolutionizing AI. Today, AI doesn’t just automate tasks—it generates new content and processes data in ways that mimic human “understanding.” With AI tools like ChatGPT, we create texts, images, and videos on demand.

Learning AI

While no prerequisites are required for our online lessons and tutorials, learning AI usually needs:

• Basic programming: Familiarity with Python and essential programming concepts.
• Math skills: Knowledge of statistics, algebra, and calculus.
• Patience: AI models are trained over time through examples, learning patterns to predict solutions.

How AI Works

Instead of manually programming every rule, modern AI models like ChatGPT learn from data. Think of training an AI as teaching it through examples. For instance, ChatGPT was trained on large datasets of text to learn patterns and then generate coherent responses.

This approach is especially useful for complex tasks, like translating languages or generating responses of text, audio, video and images. Manually programming every possible language rule is not efficient. Instead, AI models “learn” the rules with exposure to real-world examples, enabling them to generate accurate outputs even in unseen situations.

The real-world examples are provided as data (image, audio, text, video, tags, animations, 2D, 3D, synthethic …).

The data is curated / cleaned / labeled / formatted by humans and software programs and become Data Sets.

The Data Sets are then used to create Trained Data Models.

AI systems then use the Trained Data Models to interact with humans and software programs.

AI Capabilities & Limitations

AI is impressive and has limits. Large Language Models (LLMs) like ChatGPT generate responses based on patterns in the data they’ve been trained on. However, they don’t “understand” concepts like humans do and can sometimes produce incorrect or nonsensical outputs (often referred to as “hallucinations”).

Practical AI: Using ChatGPT

A simple, practical way to experience AI in action is by using ChatGPT for text-to-text tasks:

• Chat with it: Ask it questions, get summaries, or request writing help.
• Analyze responses: Does it sound like a person? Is the output accurate?

What’s Next?

In future lessons, we’ll explore more AI applications, starting with machine learning and understanding how AI models are trained. Today, just enjoy the start of your AI journey!

Lesson 1 Exercise 1:
Try out the OpenAI Playground or ChatGPT. Play around with it, and see how it responds to different prompts!

In Lesson 2, we will look at Machine Learning.

Happy learning! 😊