A Tech-Savvy Tuesday!

If you’ve ever found yourself in a shouting match with a voice assistant that just won't understand a name, or stuck in an endless loop with a website "helper" that can't grasp your question, you are in very good company. Many of us have felt the immense frustration of trying to interact with technology that is supposed to be "smart" but feels decidedly unintelligent.

For the past decade, our main experience with "talking" to computers has been through tools like Siri, Alexa, those pop-up customer service chatbots or a cable company “support” line. While those have often been referred to as using AI, those older assistants are fundamentally different from the modern AI that is making waves today. Let's explore the differences, and you have to experience it to believe it!

Modern AI vs. The Chatbots of Yesterday

Why Yesterday's "Assistants" Often Fell Short

The frustration many of us have felt with older chatbots and voice assistants is completely justified. For the most part, they weren't truly "intelligent" in the way we think of it. They typically operated on two simple principles:

1. Keyword Matching: They were programmed to listen for specific words. If you said the exact keyword, you'd get a pre-programmed response. If you phrased your question slightly differently, the system would get confused.

2. Rigid Scripts: Many customer service chatbots were built on simple decision trees. As one of our readers aptly put it, they "just pick up key phrases, and that's what you're locked into, and it puts you in this loop". They couldn't handle questions outside their script and had no ability to understand context or remember what you had said just moments before.

A Different Beast Entirely: Modern AI (LLMs)

Today's most advanced AI tools, like ChatGPT, are built on a completely different foundation called Large Language Models (LLMs). Instead of following a rigid script, these models have learned from an enormous amount of text and information from books, articles, and websites. This allows them to:

• Understand Context and Nuance: They don't just look for keywords; they understand the relationships between words and the overall meaning of your questions.

• Have a Coherent Conversation: A modern AI can remember what you said earlier in the chat and use that context to inform its next response, leading to a much more natural, back-and-forth dialogue.

• Handle Complex Ideas: They can process complicated requests, synthesize information from multiple sources, and explain complex topics in simple terms.

The Practical Difference: From Frustration to Function

This technological leap creates a world of difference in user experience.

• Instead of being stuck in a chatbot loop with your bank, imagine asking an AI to help you understand a complex medical report in plain English, and getting a clear, accurate summary back, as one of our readers, successfully did.

• Instead of a voice assistant misunderstanding a simple command, picture asking an AI to plan a multi-day itinerary for a trip to Europe, complete with hotel and restaurant ideas tailored to your specific tastes, an experience another reader found incredibly useful.

If you’re curious and want to try out AI, but don’t know where to start, send me an email. I’ll help you get set up to chat with AI without having to download a new app or create a new account.

You can reply to this newsletter or email me at steve@gestalt.com.

Quick Tech Tip

When talking to a modern AI like ChatGPT, giving it a "persona" at the beginning of your request gets better results! For example, you can start your query with, "Act as a friendly travel agent..." or "You are an expert chef. What can I make with..." This helps the AI understand the context, role, and tone you're looking for in its response. Isn’t that wild??

Tech Term Demystified: ChatGPT

You’ve likely heard the name ChatGPT in the news or from friends and family; it's the name of the specific AI product from the company OpenAI that brought this new kind of conversational AI to public attention. The name itself can be broken into two parts. The "Chat" part is simple: it refers to the fact that you interact with the tool by having a conversation or "chat" with it in plain English. The second part, "GPT," is where the technical terms come in, but they're not as complicated as they sound. GPT stands for Generative Pre-trained Transformer.

Let's quickly untangle what those three words actually mean, because they beautifully describe what the AI does. "Generative" means the AI generates or creates new, original sentences and paragraphs in response to your questions, rather than just pulling a pre-written answer from a database. "Pre-trained" means that before it was released, the AI was trained on a massive amount of text from the internet and digital books, which is where it got its vast general knowledge. Finally, "Transformer" is simply the name for the specific type of AI architecture, or 'blueprint,' that is exceptionally good at understanding the context and relationships between words in a sentence. This Transformer architecture is the key breakthrough that allows it to have a coherent, natural-feeling conversation, making it so different from the clunky chatbots of the past.

Good News Byte

AI is helping to preserve cultural heritage! Researchers are using AI models to translate and document endangered languages that have very few remaining speakers. By analyzing written texts and recordings, AI can help create dictionaries, grammars, and learning tools, ensuring this precious knowledge isn't lost to time. It's a wonderful example of using advanced tech to safeguard human history.

Did You Know?

One of the most famous early chatbots was named ELIZA, created at MIT in the 1960s. ELIZA cleverly mimicked a psychotherapist by recognizing keywords in a user's sentences and rephrasing them as questions (e.g., if you wrote "I'm sad," it might respond, "Why are you sad?"). While it was just a simple script, some users were famously convinced they were talking to a real person, showing that our desire to have conversations with technology has been around for a very long time!

Your Turn to Experience the Difference!

Feeling curious? Send me an email and will help set you up to try chatting with one of these Large Language models, right from your email.

Email: steve@gestalt.com

A Little Bit of Fun

Talking to an old chatbot is like talking to someone who only listens for one specific word in your sentence. Talking to a modern AI is like talking to someone who has read the entire library... and actually remembers what's in the books!

Wishing you a week of clear communication!

Warmly,

Steve