Artificial Intelligence

What About AI?

Yeah, what about it?

Well... is AI taking over completely? Are human beings now obsolete? Will AI take all of our jobs away? And most importantly, is there any compelling reason to learn web development when AI can do it much faster and better? And are we all going to die when the robot overlords realize that humans are vastly inferior, and that all of us must be summarily eliminated?

These are all great questions to ponder, especially if you have aspirations of becoming a science fiction writer. Or do you think that AI will take their jobs away as well? It was only about a month ago when we felt that such apocalyptic projections were largely unwarranted. Recent events, however, have caused us to rethink some of our assumptions about where AI is headed.

TL;DR

Click here to get to The Bottom Line.

For hundreds or even thousands of years, the human race has struggled to adapt whenever some new innovation or technology has disrupted the old ways of doing things. Imagine the skilled craftsmen who were proficient at making arrowheads and spearheads from flint, obsidian, and quartz when they first encountered similar objects made from bronze. Some of them probably wondered what this new technology meant for their future. Those who learned to work with the new materials and techniques of the Bronze Age would have acquired valuable new skills in metallurgy. Meanwhile, the old technology did not simply disappear overnight. Stone and metal tools existed side by side as societies gradually adapted to this new metallic world. And we can safely assume that somebody, somewhere, was probably complaining about the new technology. Some things never change.

And there are literally countless similar scenarios we could reenact here. What happened to all of the scribes who used to produce beautifully hand-crafted books using the fine art of calligraphy when Gutenberg introduced printing presses with movable type? Did all of the post offices close when email was first introduced? How did the telephone companies survive the cellular mobile phone industry? Surely AT&T had to lay down and die when that happened, right? There was a time not long ago when factory workers worried that automation would eventually take all of their jobs away. Did that happen? No. Automation certainly eliminated some jobs and changed many others, but manufacturing itself did not disappear. Instead, technology, international trade, and changing global supply chains transformed where products were made, how they were made, and what skills workers needed to make them.

We all know that technological change can be disruptive. Those who can embrace the possibilities of the future and learn to adapt to these changes will generally be in a better position to take advantage of them. Those who stubbornly resist every change simply because it is new may find themselves left behind. And this is the case with AI as well. Of course, we fully understand that adapting to constant change isn't always easy. But some things never change, and the ability to adapt will remain enormously important with each new technology that comes along.

Which AI?

Yes, it's a jungle of a market. Every possible large corporation seems to be pushing AI in your face right now. And many of the names of AI products and the companies behind them might be unfamiliar to some of you. So let's take it slow here. Let's talk about what AI actually is, and how it is being served to us, before we talk about actual products and companies.

When we talk about this murky subject of Artificial Intelligence, it can get very deep fast. There's a ton of new terminology mixed in with even more misinformation. It can sometimes seem difficult to separate the grain from the chaff, or the flecks of gold from the bovine excrement. What we will attempt to do here is to boil this mess down into sensible terms that almost anyone can understand. Let's start with the simplest of terms.

Chatbots

Chatbots are software applications designed to simulate human conversation through text or voice. They accept user input and respond automatically, allowing people to interact with digital services through something resembling a human conversation. Have you ever used an online customer service chatbot or a similar voice-synthesized chatbot on your phone? They say, "You can ask questions like, 'What is my balance?' Or 'How can I pay my bill?' What was that? I'm sorry. I did not understand your question." And if you are successful at getting the chatbot to put you in a queue to speak to a real person, you are usually on hold while recordings tell you "Your call is important to us. Please hold for the next available assistant. Your wait time is approximately ... [in a different voice] ... 42 minutes."

If you've played this little corporate game, then you already understand one of the limitations of traditional rule-based chatbots. They don't necessarily understand what you're saying. They generally follow predefined rules, recognize certain patterns or intentions, and respond accordingly. Go too far outside what they were designed to handle, and things can fall apart pretty quickly.

AI Chatbots

And although all chatbots are meant to be interactive, AI Chatbots are very different from traditional rule-based chatbots that rely on predefined rules, scripts, and menu options. Modern AI chatbots like ChatGPT, Claude, and Gemini use Large Language Models (LLMs) to generate dynamic responses based on what we ask them and the context of the conversation.

Instead of simply choosing from a list of predetermined answers, an LLM generates its response as it goes. At its most basic level, it repeatedly predicts what pieces of language, called tokens, are most likely to come next. That might not sound very intelligent, but when this process is performed by enormous neural networks trained on vast amounts of text and other data, the results can be astonishingly sophisticated.

This is why modern AI chatbots seem so different from the frustrating rule-based chatbots we've been dealing with for years. They can explain complicated subjects, summarize information, translate languages, write software code, recognize patterns, and carry on conversations that can seem remarkably human. But don't let that apparent intelligence fool you. An AI chatbot does not necessarily understand something in the same way that a human being understands it. And as we will soon discover, sounding intelligent and actually being correct are two very different things.

Understanding Context

One of the things that makes modern AI chatbots seem so intelligent is their ability to understand the context of a conversation. You don't necessarily have to phrase every question perfectly or use specific keywords. You can use slang, abbreviations, incomplete sentences, humor, and sometimes even metaphors, and a good AI chatbot can often figure out what you mean.

More importantly, the chatbot can use information from earlier in the conversation when responding to something you say later. If you're discussing JavaScript and then ask, "Why isn't this working?", you probably don't need to explain all over again that you're talking about JavaScript. The chatbot can use the surrounding conversation to help determine what "this" means.

This ability can make a conversation with AI seem remarkably human. The chatbot may also adapt its writing style to the way you communicate with it. If you're casual and humorous, it may respond in a similar style. If you ask for short and direct answers, it can usually accommodate that preference. If you want detailed technical explanations, it can do that instead.

Some AI products also have memory or personalization features that can retain certain information and preferences between conversations. That's different from simply remembering what was said earlier in the current conversation, and these capabilities vary considerably between AI products.

But be careful about assigning too many human qualities to what is happening. The chatbot isn't necessarily studying your personality, forming opinions about you, or developing a new personality of its own. It is responding to your instructions, the context available to it, patterns it learned during training, and any personalization features provided by the particular AI product.

And this leads to an important distinction that we'll come back to later: an AI chatbot adapting its responses during a conversation does not mean that the underlying AI model is being retrained or permanently learning from everything you tell it.

GPT

We keep throwing around the term GPT, so let's make sure we understand what it means. GPT stands for Generative Pre-trained Transformer. That sounds sufficiently intimidating, but the three words aren't nearly as frightening when we take them apart.

Generative means that the AI can generate new content. Instead of simply choosing from a collection of predetermined answers, it can generate sentences, explanations, summaries, software code, and many other forms of output in response to what we give it.

Pre-trained means that the AI model went through an enormous training process before we ever started talking to it. During that training, it learned statistical patterns and relationships from huge amounts of data. By the time we encounter the finished model in an AI chatbot, much of the heavy lifting has already been done.

And then we have the scary word: Transformer.

A Transformer is a type of neural-network architecture that is particularly good at recognizing relationships and patterns within sequences of information such as language. One of its important features is called attention, which helps the model determine which parts of the information it is processing are most relevant to other parts.

For example, consider the sentence, The programmer couldn't fix the computer because it wouldn't start. We immediately understand that it probably refers to the computer rather than the programmer. A Transformer uses learned relationships and context to make these kinds of connections while processing language.

You don't need to understand the mathematics behind Transformers to use AI effectively. What matters here is that the Transformer architecture is one of the technologies that made today's remarkably capable Large Language Models possible.

So when you hear GPT, don't let the acronym scare you. We're simply talking about a model that is Generative, has been Pre-trained, and is built using a Transformer architecture.

LLMs

We've already used the term LLM several times, so let's take that one apart as well. LLM stands for Large Language Model.

The Large part refers to the enormous scale of these models. Modern LLMs can contain billions of adjustable numerical values called parameters, and training them requires enormous amounts of data and computing power.

The Language part tells us what they were originally designed to become very good at: recognizing and generating patterns in human language. Modern models can work with much more than ordinary text, but language remains central to what we mean when we call something an LLM.

And the Model is the resulting mathematical system that has learned all of those patterns during training.

LLMs are built using neural networks. Despite the name, don't imagine a little electronic human brain inside your computer. Artificial neural networks were inspired, very loosely, by ideas about interconnected biological neurons, but they are mathematical and computational systems rather than artificial brains.

This brings us to an important misconception. An LLM is not simply a gigantic database containing everything the AI knows. During training, the model adjusts enormous numbers of internal parameters as it learns statistical patterns and relationships from its training data. Those learned relationships allow it to generate remarkably useful responses to questions it has never encountered in exactly the same form before.

That also helps explain one of the strangest characteristics of AI chatbots. An LLM can produce an answer that sounds extremely confident and convincing without retrieving that answer from a verified database of facts. Sometimes the answer is correct. Sometimes it isn't. This is one reason why you should never automatically assume that an AI-generated answer is true simply because it sounds intelligent.

Modern AI applications can also be given additional capabilities. They may search the web, retrieve information from databases, examine documents, run software tools, or access other sources of current information. But those capabilities should not be confused with the LLM itself. The LLM is the underlying model that processes the information and generates the language we interact with.

Machine Learning

Another term that gets thrown around constantly when discussing AI is Machine Learning. Fortunately, the basic idea isn't nearly as complicated as it sounds.

Traditional computer programs are generally created by programmers who explicitly write instructions telling the computer what to do. With Machine Learning, developers instead use data and training techniques that allow a computational model to learn patterns and relationships. During this training process, the model's internal numerical parameters are repeatedly adjusted so that its results become better at accomplishing the task it is being trained to perform.

This distinction can become confusing when we're using an AI chatbot because the chatbot can appear to be learning from us while we're talking to it.

For example, suppose you ask an AI chatbot a question and it gives you an answer that you know is wrong. You challenge the answer and provide additional information. The chatbot considers this new information and responds, "You're right. I apologize for the mistake." It may then give you a much better answer.

Did you just teach the AI something through Machine Learning?

Not necessarily.

During an ordinary conversation, your correction becomes additional context that the AI can use when generating its next response. The underlying model isn't necessarily being retrained or having its billions of parameters permanently changed simply because you corrected one of its answers. It is using the additional information you provided to reconsider the question.

Some AI products also have memory or personalization features that can retain certain information between conversations. That's useful, but it still isn't the same thing as retraining the underlying model.

And user feedback can sometimes contribute to the development and improvement of future AI models, depending on the particular product, its policies, and the user's settings. But that improvement happens through separate training and development processes rather than because the chatbot instantly rewrites itself every time somebody catches it making a mistake.

So when we talk about Machine Learning, we're talking about the broader process through which these systems learn patterns from data during training. When an AI chatbot changes its answer because you gave it more information during a conversation, that's usually better understood as the model using context, not as the model suddenly going back to school.

Generative AI

Now we finally get to one of those terms that actually means pretty much what it sounds like. Generative AI refers to artificial intelligence systems that can generate new content in response to instructions or other input.

That content can include text, software code, images, audio, video, and other forms of digital information, depending on the capabilities of the particular AI system.

When you ask an AI chatbot to write a paragraph, summarize an article, explain some JavaScript, or create a block of HTML and CSS, you are using Generative AI. When you describe a picture and an AI system creates an image from that description, that's Generative AI as well.

The word generate is important. The AI isn't necessarily searching through a giant database looking for a finished paragraph, image, or computer program that matches your request. Instead, it uses patterns learned during training, together with your instructions and other available context, to generate an appropriate output.

This is one of the reasons Generative AI has become so useful to web developers. You can describe what you're trying to accomplish in ordinary language and ask the AI to help you write, explain, debug, or improve your code. You can also use generative AI to help create some of the text and visual content that goes into a website.

But there's an important word in that last paragraph: help.

Generative AI can produce remarkably good results, but it can also produce code that doesn't work, explanations that are misleading, or information that is simply wrong. The more you understand about the subject yourself, the better equipped you are to recognize the difference.

And that's one of the main reasons why you're here learning web development in the first place!

Agentic AI

And now things get really interesting.

An ordinary AI chatbot generally waits for you to ask a question or give it an instruction and then generates a response. Agentic AI goes considerably further. An AI agent can be given a goal and some degree of autonomy to take a series of actions in an attempt to accomplish that goal.

For example, instead of asking an AI chatbot, What's wrong with this JavaScript function?, a software developer might give a coding agent access to an entire project and ask it to fix a particular problem. The agent can examine multiple files, determine which ones may need to be changed, edit the code, run tests, examine the results, and continue working through the problem.

That's a fundamentally different relationship with AI. Instead of merely asking AI for advice, we're allowing AI to do some of the work.

AI Chatbots Versus Agentic AI

So what's the practical difference between using an AI Chatbot and using Agentic AI?

Suppose you're working on a website and something isn't working properly. You could copy a section of your code into an AI chatbot and ask, “What's wrong with this?” The chatbot can examine the code you provided, explain what it thinks the problem is, and suggest changes that might fix it. But you're still sitting in the driver's seat. You decide which code to show it, you decide whether its suggestions make sense, and you make the changes to your website.

An AI coding agent can take that relationship considerably further. If you give the agent access to your project, you might simply tell it what isn't working and ask it to fix the problem. The agent can examine the files in the project, determine which ones are relevant, make changes to the code, run available tests or other development tools, examine the results, and continue working toward the goal you gave it.

The difference isn't that one of these systems is intelligent and the other one isn't. The important difference is the amount of autonomy we're giving the AI. With a chatbot, we're generally asking AI to tell us what we should do. With an agent, we're allowing AI to take some of those actions for us.

And there is plenty of territory between those two extremes. An AI system might suggest a change and wait for your approval before making it, or it might be given permission to perform certain actions automatically while requiring approval for others. How much autonomy an agent has depends on the particular product, the tools it can access, and the permissions we give it.

Force Feeding

---

WARNING: The force feedings will continue until morale improves. Open your mouths wide. We need to shove Copilot and Gemini down your throats as forcefully as possible. Don't you realize that AI is the future? Don't you clamor for its help? We will continue this practice of handcuffing you behind your back and helping you, whether you require our assistance or not.

---

Interesting enough, search engines have vastly improved over the past few years. You can ask your search engine almost anything and generally you will get good reliable answers back. We are certain that has everything to do with implementing the tools of AI into search engine algorithms. And it certainly is beneficial to have these tools at our fingertips. But your participation should always be optional.

If you've read this far, then you probably understand that we're not opposed to AI. Quite the opposite. AI has already become an incredibly useful tool for researching information, explaining difficult subjects, writing and debugging code, and helping us find answers that used to require digging through countless web pages and support documents.

What we aren't quite as enthusiastic about is having AI features forced into every piece of software we use, whether we asked for them or not.

There's an important distinction here. We want to use AI when it makes something easier or better. We don't necessarily want every operating system, web browser, email program, PDF reader, and other application constantly encouraging us to use its own built-in AI assistant. And when these features are enabled by default, prominently displayed, or difficult to remove, some users may reasonably wonder whether the software is serving them, or whether they're being asked to serve the marketing ambitions of the company that made it.

Ironically, one of the things AI has become particularly good at is helping us figure out how to disable unwanted AI features. Instead of spending half an hour digging through support documents and application settings, we can ask an AI chatbot, "How do I turn this thing off?" And quite often, it will point us in the right direction.

So our objection isn't to AI itself. It's to lack of choice. Give us useful AI tools, show us what they can do, and then let us decide whether we want to use them.

AI Market Share

So which AI chatbot should you use?

There certainly isn't any shortage of choices, and there is nothing preventing you from trying several of them and deciding which one works best for you. But if you're completely new to AI and simply want to know where we recommend starting, our recommendation is ChatGPT.

And we're certainly not alone. According to Statcounter GlobalStats, ChatGPT accounted for about 79% of its measured worldwide AI chatbot market share in August 2026. Google Gemini was a distant second at about 11%, followed by several smaller competitors. These numbers change constantly, so don't become too attached to the exact percentages. Follow the link to Statcounter if you want to see where they stand today.

But market share isn't why we recommend ChatGPT. The most popular product isn't automatically the best product for everyone. We recommend ChatGPT because of our own experiences using it, the quality of the conversations we've had with it, and the broad range of things we've found it useful for. Your experiences and preferences may be completely different.

And remember the distinction we made earlier between chatbots and agents. If you're just beginning to use AI while learning web development, we think a general-purpose chatbot is an excellent place to start. Ask it questions. Ask it to explain code you don't understand. Ask why something isn't working. Ask it to compare different ways of solving a problem. And yes, let it help you write some code.

But learn enough about that code to understand what you're asking for and to recognize when the answer doesn't make sense. You can always explore increasingly powerful AI coding agents as your knowledge and experiences with AI grow.

The Bottom Line

So... is AI taking over completely? And will humans eventually be banished from the workforce? We don't pretend to know exactly where all of this is headed. What we do know is that AI is here to stay, and it will almost certainly change the way many of us work. Some jobs will disappear, some will change considerably, and entirely new jobs will probably emerge along the way. We've been through technological disruptions before, although this one certainly seems to be moving extraordinarily fast.

But what about the future of Web Development? Never lose sight of the fact that AI chatbots and AI agents are tools. And like any other tools, their usefulness depends considerably upon the knowledge and skills of the people using them. AI can already generate impressive amounts of software code, but somebody still needs to decide what the software is supposed to do, determine whether the results actually work, test them, recognize problems, and make intelligent decisions about what happens next.

That's why we keep repeating our mantra that there are no shortcuts. The more you understand about the basics of web development, the better equipped you will be to use these increasingly powerful AI tools. You'll be better able to ask intelligent questions, recognize bad answers, test the code AI produces, and understand why something works — or why it doesn't.

And lastly, we'll offer this little pearl of wisdom. While you're learning, we think there is no better coach than a good AI chatbot. We recommend ChatGPT, but there are several excellent AI chatbots available, and you're certainly free to decide which one works best for you.

We also recommend that beginning web development students don't become too dependent upon AI coding agents too quickly. An agent may be able to examine your project, write the code, modify files, test its work, and accomplish in minutes what might take you hours to do yourself. That's enormously useful when getting the job done is your primary objective. But when learning is the objective, those frustrating hours of figuring things out can be enormously valuable.

So ask your AI chatbot questions. Ask it to explain things you don't understand. Ask it why your code doesn't work. Ask it to show you different ways of solving the same problem. And yes, let it generate some code for you. But read the code. Ask questions about it. Change it. Break it. Fix it. Learn from it.

Use ChatGPT as your coach, rather than asking AI to run the entire race for you. That's a great way to learn the basics of web development. And if you need some help getting started, that's exactly what we're trying to accomplish with our classes here at AWD. We admit openly that ChatGPT can explain some things better than we can. But we can help you learn the fundamentals you'll need before you know which questions to ask.

Yes, AI is here to stay. But so are you, if you learn how best to utilize these extremely powerful AI tools. And if you still don't believe that web development can be a lucrative career choice, then we strongly recommend a career in tattoo removal. We have a feeling that particular profession will remain in demand for a very long time.

The Next Blog