Artificial Intelligence
Artificial intelligence (AI) is a technology used to create computer systems capable of human-like intelligence. You’ve already experienced AI through automated phone calls, virtual assistants, and website chatbots. In this section, we’ll look at the basics of artificial intelligence to better understand how conversational AI works.
In this lesson:
- What is artificial intelligence?
- How does AI work?
- Conversational AI examples
What is artificial intelligence?
Artificial intelligence (AI) is a new technology that seeks to imitate human intelligence. Although computers don’t have feelings or emotional intelligence, computers can analyze vast amounts of data quickly. This ability is useful in many ways, like fraud detection, medical diagnosis, and weather prediction.
Within the realm of AI, there are specialized fields, each focusing on different aspects of human intelligence. One such field is machine learning (ML), where computers are trained to analyze vast amounts of data.
Another related field is natural language processing (NLP), which seeks to imitate human speech, including understanding sentiment and emotion. NLP achieves this by predicting likely responses based on its extensive dataset.
Together, these fields combine to build conversational AI models like ChatGPT, Gemini (Google), and Copilot (Microsoft).
How does AI work?
Advances in computing power allow for new programming methods. Rather than providing the computer with explicit instructions, AI developers provide raw data and let the computer detect the pattern. This capability can then be scaled up to analyze very large sets of data.
Suppose you want to train an AI to detect possible fraudulent activity. First, you provide the AI with your labeled data, including good transactions and bad ones. This allows the system to learn by detecting patterns in the data. Second, you provide additional data, but unlabeled. The system flags the transactions it thinks are fraudulent. This allows you to test the results. Third, you’ll fine-tune your algorithms if needed. This is an ongoing process as current AI models are under constant development.
- Train on known data
- Test on unknown data
- Fine-tune
As mentioned earlier, the advantage of the “machine learning” method is that an AI model can analyze huge amounts of data based on pattern detection, which is applicable in many fields. In this course, we’ll talk about conversational AI for beginners.
Prompt engineering
Conversing with AI is often called “prompt engineering,” an iterative, turn-based process that involves:
- Prompt Engineering: Crafting questions or prompts to extract specific information or responses from the AI
- Prompt Tuning: Adjusting your questions or prompts based on the AI’s output to refine the conversation further
Examples
Large language models such as ChatGPT are trained on extensive datasets and have the ability to communicate in simple terms. Mastering conversational AI involves asking the right questions to elicit the desired responses from the AI system.
Example:
[User]: “Let’s talk like pirates.”
[Chat]: “Shiver me timbers!”
[User]: “This be fun!”
[Chat]: “Yar!”
In the above example, the user issues an instruction (prompt) and then provides positive feedback.
Example:
[User]: “Any good movies?”
[Chat]: “There’s a Pacino flick at the Royal.”
[User]: “Read me some reviews.”
In this interaction, the AI recognizes the context of the conversation and provides relevant information about the Pacino movie.
What’s next?
In this lesson, we looked briefly at artificial intelligence (AI). In the next lesson, we’ll look at how AI systems are trained on data.