By the end of this module, you will have a solid understanding of what AI is, how it works, and some of its potential applications. We’ll review a brief history of AI, so you know where it comes from, as well as some key concepts like machine learning, neural networks, and natural language processing.
‍
From there, you will also learn about the process of creating AI systems, including data collection, algorithm design, and model training. This is important for understanding future concepts such as data privacy and bias.
‍
Lastly, to connect all of this to current events, we’ll review some of the key organizations at the center of contributing to the development of AI, such as OpenAI, Anthropic, Google DeepMind, and Meta.

Use this section like flashcards

Artificial intelligence (AI) is a type of software that differs from other kinds of software in several key ways. Unlike traditional software, which has a predetermined set of instructions, AI can learn and adapt to new situations, making it more flexible and versatile.

For example, when you think about a calculator app on your phone, it can perform calculations, but it can't do anything else. It can't learn to do something new, like recognize your handwriting or understand spoken commands. AI, on the other hand, can be trained to recognize images, understand human speech, and even play games like chess or Go.

Another key difference between AI and other kinds of software is that AI can make decisions based on incomplete or ambiguous information. Traditional software requires clear instructions and fixed rules, but AI can make educated guesses and predictions based on patterns in data. This makes AI useful for tasks like fraud detection, where it can identify suspicious behavior even when the evidence is not straightforward.

Finally, AI is designed to be self-improving. As it is exposed to more data and experiences, it can learn from its mistakes and improve its performance. Traditional software requires updates and patches to fix bugs and add new features, but AI can improve on its own, making it a powerful tool for a wide range of applications.

World chess champion Garry Kasparov playing against Deep Blue, the chess-playing computer built by IBM. In 1996 Kasparov won the first match 4−2, but in 1997 he lost to Deep Blue 3 ½−2 ½.

In the mid-twentieth century, computer scientists started exploring the idea of creating machines that could simulate human intelligence. At that time, computers were still relatively new and could only perform simple calculations and store data.

One of the earliest strategies for creating AI was to build expert systems, which were designed to mimic the decision-making abilities of human experts in specific domains. They used a set of rules and heuristics to make decisions based on a given set of inputs.

Deep Blue, developed by IBM in the mid-1990s, is a noteworthy example of an expert AI system. It was a chess-playing computer that defeated a reigning world chess champion, Garry Kasparov, in a six-game match in 1997. Deep Blue used a brute-force approach to analyze millions of possible moves per second, along with a database of past games to make decisions. Its victory over Kasparov marked a major milestone in the field of AI.

One important milestone in the history of AI was the development of neural networks, which were inspired by the structure and function of the human brain. Neural networks consist of interconnected nodes that process and transmit information, and they can learn from experience to improve their performance on a given task.

In the past few years, AI has become incredibly popular thanks to advancements in machine learning and natural language processing. These breakthroughs have led to the development of powerful new applications like speech recognition, image recognition, and self-driving cars.

But as AI continues to evolve and become more sophisticated, it also raises important ethical questions about the role of machines in our lives and society. It's important to consider the impact that AI may have on us as we move forward.

The process of creating AI systems involves several key steps, including data collection, algorithm design, and model training.

To create an AI system, you need data. Lots of data. This data can come from a variety of sources, such as sensors that collect information about the environment, social media feeds that capture human behavior, or scientific studies that provide insights into how the world works.

It is important to ensure that the data you collect is representative of the problem you are trying to solve. If your data is biased or incomplete, your AI system may not perform well or may even make harmful decisions.

The word "algorithm" comes from the name of the Persian mathematician Al-Khwarizmi, who was one of the first scholars to write about algebraic methods.

An imaginary portrait for Al-Khwarizmi derived from a soviet stamp.

An AI algorithm can be designed using a variety of techniques, including rule-based programming, statistical models, and machine learning algorithms.

Rule-based programming involves creating a set of if-then statements that the AI system can use to make decisions based on the input data.

Statistical models involve using mathematical models to analyze the data and make predictions based on patterns in the data.

Machine learning algorithms involve training the AI system on a large amount of data and adjusting the algorithm's parameters until it can accurately predict the correct answers.

Overall, the algorithm is the heart of any AI system, and the design of the algorithm is a critical step in the process of creating AI systems that can learn from data and make predictions based on that learning.

Once you have your data and your algorithm, you need to train your model. Model training involves feeding your algorithm with data that has been manually sorted or annotated, and allowing time for the model to begin learning patterns and probabilities inherent within the data set.
​

Professional model, Derek Zoolander, training

Training an AI model is similar to teaching a child to identify, say, animals. With enough examples and guidance, the child gains the ability to differentiate between a cat and a dog.

Likewise, an AI model is trained with data, after which it is tested and provided feedback by human software engineers, allowing it to improve its prediction accuracy. The more data the model sees, the better it gets at recognizing patterns and making predictions.

Once a model is trained, you can use it to make predictions on new, unlabeled data. This is the essence of AI: creating systems that can learn from data and make predictions based on that learning.

This short quiz is only meant to help you check your understanding of these materials. Your score is not recorded, so please write it down if you want to keep track for your own records.