Types of AI

AI is all around us, and you might already be able to recognize some of the most common types of artificial intelligence.

• Natural Language Processing (NLP): Search engines use NLP algorithms to understand the meaning of searches and match them with relevant results. The AI uses context clues and the way you have written your search to figure out what you may be looking for. For example, when you ask Google or Siri something, the computer processes your voice and connects that to words to understand what you are looking for.
• Machine Learning (ML): Search engines use ML algorithms to learn from your behavior and to personalize search results. For example, if you frequently click on articles from a particular news source, the search engine may put results from that source at the top in future searches.
• Natural Language Generation (NLG): Some search engines use NLG algorithms to generate summaries search results. This can help users quickly understand the relevance of a particular search result without having to read the entire article.
• Image Recognition: Search engines use image recognition technology to look at images and match them with relevant searches. This is particularly useful for image-based searches, such as searching for a specific type of product or landmark. For example the iNaturalist app could take a picture of a plant and tell you what it is based on the images in the database.

You can easily test natural language processing AI using your browser.

1. Write a sentence down on a piece of paper.
2. Open google.com or another search engine in a new tab.
3. Start typing the first word into the search engine. What is the suggested second word? Is it your word or a different word?
4. Now type in the second word of your sentence. What is the third word that is suggested in the search engine?  Is it your word or is it further down the list?
5. See how many words you have to type into the search engine before it can correctly start predicting the next word.
6. Try a different sentence, but this time find a sentence that is in a book you have. Does it predict the words easier?
7. Now try the same activity using a text message and see if the text message suggestion can predict the next word in your phrase or sentence.

This AI relies on certain words usually following other words. Machine-based learning, another type of AI, also plays a role in this. When you use a suggested search, it uses its accurate predictions to provide better search results next time.

Did you know?

Did You Know? Ada Lovelace was a 19th-century British mathematician and writer who is often credited as the world’s first computer programmer. When she was 28, Ada was introduced to Charles Babbage, an inventor who had designed a machine called the Analytical Engine, which could perform complex mathematical calculations. Ada’s work on the Analytical Engine was groundbreaking. She was the first person to recognize that the machine had the potential to do more than just perform calculations. She thought the Analytical Engine could be used to manipulate symbols and represent complex data, which is a key concept in modern computing.

AI and Virtual Assistants

Natural Language Processing (NLP) is a way for computers to understand and work with human language, just like we do when we talk or write. It helps computers understand what we are saying and respond in a way that makes sense.

Think of it like this: You know how we can read books and understand the words and sentences? Well, NLP helps computers do something similar. It helps them read and understand text or listen and understand speech.

With NLP, computers can:

• Help us find information on the internet by understanding what we’re looking for
• Translate words and sentences from one language to another
• Help us chat with virtual assistants like Siri or Alexa

AI is a key technology used in virtual assistants. Some common virtual assistants are:

• Google Assistant
• Siri
• Alexa
• Cortana
• Bixby

This doesn’t cover all of the virtual assistants out there, but hopefully it gives you a good idea of what we mean by “virtual assistant.”

Virtual assistants answer questions, provide information, and control smart home devices, but there is a lot of technology behind all these useful abilities.

• Speech Recognition: Virtual assistants use speech recognition technology to convert spoken words into text. This enables the assistant to understand what the user is saying and to respond appropriately.
• Query or Question Processing: Once the assistant has understood the user’s query (a fancy word for question), it processes the information to determine the best response. This may involve accessing a database of information, performing a web search, or executing a task such as setting an alarm or sending a message.
• Response Generation: After processing the user’s query, the virtual assistant generates a response in natural language, which is then spoken or displayed to the user. The assistant may also perform a task, such as sending an email or controlling a smart home device, in response to the user’s query.
• Learned Behavior Patterns: Virtual assistants use machine learning algorithms to learn from user behavior and personalize responses. For example, if a user frequently asks for weather updates in a particular location, the assistant will learn to prioritize weather information for that location.

Even though it may seem like a virtual assistant knows everything, they only know what they’ve been programmed with. They do not think on their own, but instead, access a set of knowledge and tasks that a human taught them. Here’s an exercise to try:

1. Ask the virtual assistant 10 questions that you know the answer to. Did the virtual assistant get it right, or did it answer, “I don’t know?”
2. Now, ask the virtual assistant harder questions you do not know the answer to.
3. Next, ask the same question on a search engine.  Were the answers different or the same?  Did each source give the same answer?

The potential of AI is vast and exciting, with the potential to transform the way we work, learn, and interact with the world around us.

Did you know?

Did You Know? AI is changing the way we live, work, and play! In the business world, it has been used to automate tasks previously performed by humans. These include customer service work, lead generation, fraud detection, and quality control. In fact, AI is often better suited for repetitive and detail-oriented tasks, such as analyzing large volumes of legal documents, where it can complete jobs quickly and with minimal errors. The power of AI lies in its ability to process a lot of data, showing us patterns we may not have noticed before. This makes it a valuable tool for businesses looking to improve their performance and stay ahead of the competition.

OpenAI and ChatGPT

Natural Language Generation (NLG) is like having a computer that can write stories or sentences all on its own.  NLG helps computers generate text that sounds natural and makes sense.

Before we get started, see if you can get together at least three people to make your own natural language generation AI (sort of). Natural language generation is an intimidating topic, but by applying yourself and working with friends, you can understand it!

Here’s what you’ll need to do:

1. Pick a story topic. (This could include “ghost story,” “adventure at sea,” “about my day,” or anything you can think up. Get creative!)
2. Decide who in the group will go first.
3. Using your topic, tell a story, one person, one sentence at a time.
4. Keep going until you’ve rotated through all the people in your group and have created a finished story.

Your finished story is an example of natural language generation! The ideas of the people in your group are like the data a computer is given. By getting ideas from each person and putting them together, you generated a story from multiple sources, just like a computer using natural language generation would do.

OpenAI is an AI research organization. Their mission is to create and promote friendly AI that benefits humanity. OpenAI is the organization behind ChatGPT, one of the newer AI tools you’ve probably heard of. ChatGPT stands for “Generative Pre-trained Transformer.”

You may have heard that ChatGPT (and other AI options) can give very human-like answers. That is because ChatGPT is based on Natural Language Generation. How does ChatGPT work?

1. Code is written. Coders at OpenAI write the code necessary for ChatGPT to run.
2. ChatGPT starts reading. ChatGPT learns how to generate text by “reading.” Coders give ChatGPT access to books, news articles, and websites to help it get used to language and make predictions about words.
3. Trainers step in. After ChatGPT has learned all it can on its own, trainers begin to add new words, sentences, and stories to help it sound more human.
4. ChatGPT answers are ranked. Now that ChatGPT has a good idea of what human conversation should sound like, AI trainers start to ask it questions, ranking the answers based on quality. From this, ChatGPT learns how to provide the best answers.

Did you know?

Did You Know? OpenAI has released several versions of GPT, with the first being released in 2018. With each new version of their GPT, OpenAI is increasing the skill level, training time, and amount of training data. ChatGPT is being used for many purposes, from having a simple conversation partner to generating content based on specific questions. Many companies are adding their own AI tools into their applications. Examples include Google Bard, Microsoft Bing, and Meta Perplexity.