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.