AI: The Invisible Power Behind the Screen

Just like flipping a switch to turn on a light, most of us do not think about what had to happen to make that light turn on. Artificial Intelligence (AI) has already become as effortless as a light switch. You type a prompt into ChatGPT or Copilot, and it returns an answer, yet you likely have no idea how it works or what went into making it function. Unlike understanding electricity or the internet, grasping AI requires a foundational knowledge of its mechanics. While the full explanation is more complex than I can cover here, let’s break down some basics.

The Rise of AI and Large Language Models (LLMs)

AI has existed for decades, but OpenAI’s introduction of an easy-to-use chatbot brought it into the mainstream. Now, the tech industry is laser-focused on AI advancements. 

OpenAI’s ChatGPT is an example of a LLM. These models are trained on massive datasets; often encompassing nearly the entire publicly available internet. Essentially, they have processed vast amounts of published data, from historical texts to modern web content. 

However, a LLM does not “think” like a human. Instead, it excels at recognizing patterns in language. Think of it as a consultant who has absorbed millions more books than any human could in a lifetime. Need a market analysis outline or a polished response to a customer query? A LLM can generate an answer based on its training.

The Heavy Lifting Behind LLMs

Training a LLM demands immense computing power. NVIDIA, originally known for gaming graphics processing units (GPUs), now plays a crucial role in AI. Modern 3D games require GPUs to perform trillions of operations per second. Similarly, AI models rely on server farms packed with high-performance GPUs to process trillions of data points.

This computing power does not come cheap. Developing and training a LLM is expensive, yet there are already over 1.5 million LLMs, with more emerging daily. You can explore many of them (including free, open-source models) on platforms like Hugging Face (https://huggingface.co/models). Some can even be downloaded and run locally—more on that in a future post.

Popular AI Chat Models and Their Specialties

This chart shows some of the most popular AI chat models

AI Chat Tool	Sponsoring Company	Specialty/Focus Area
ChatGPT	OpenAI	General-purpose conversational AI, writing tasks, coding, and creative content generation
Microsoft Copilot	Microsoft	Productivity enhancement, integration with Microsoft Office tools, and business support
Claude	Anthropic	Privacy-focused AI chatbot designed for secure and confidential conversations
Perplexity.ai	Perplexity AI	Specialized in search and information retrieval through conversational AI
Jasper	Jasper AI	AI chatbot for businesses and marketers, content creation, and editing
Llama	Meta Platforms	Open-source AI model designed for natural language understanding and generation
Grok	X	AI assistant with advanced reasoning, coding, and visual processing capabilities
Deep Seek	Deep Seek (China)	Free; Giving best results as of March 20, 2025

Each model has its strengths, so experimenting with different tools can help you find the best fit for your needs. 

The Future of AI Chatbots 

Competition in the AI space is fierce, and models will only improve with each iteration. In my next post, I will cover prompt engineering techniques to help you get better AI responses—stay tuned!

Want to learn more about LLMs? Check out my five-minute YouTube explainer: https://www.youtube.com/watch?v=O27yoZhq9dw

Mark your calendar! NACVA is hosting its AI Launch Webinar April 30, 2025, where we will introduce upcoming AI training sessions designed to help you integrate AI into your work. 

Visit the AI Data University at https://www.aidatauniversity.com/ to learn more and search for AI resources.

---

Colin Brown, can be contacted at https://syncnet.com or by e-mail to cto@syncnet.com.