Use of Artificial Intelligence in Web Design

Nothing like asking the horse’s mouth.

“Who owns the copyright of the responses you write? Do you ever present content written by others, and still in copyright, as if you had written it yourself?”

“The responses I write as ChatGPT are generated using machine learning algorithms based on patterns and information available in my training data. As a result, the copyright of the responses I generate belongs to OpenAI, the company that developed and operates the GPT-3.5 language model.

“I do not present content written by others as if I had written it myself, as I am programmed to generate original responses based on the input provided to me. However, if a user were to provide me with content that is protected by copyright, it would be the responsibility of the user to ensure that they have the appropriate rights to share that content with me before I generate a response based on it.”

Good idea to rewrite, then.

Incidentally, we‘re focusing a lot of attention on the generation of content here. And there’s no doubt that Chat GPT and other LLMs can do amazing things with content. Nonetheless, to come back to the other things we’re concerned with in this forum, bear in mind that Chat GPT can write very efficient, competent HTML, CSS and Javascript (I’ve not tried PHP, so I don‘t know about that), and it also explains what the code does and how it is to be used. What‘s more, you can iterate it over and over again until you get exactly what you want.

I know I’ve mentioned this before, but it’s worth reiterating. Just a little bit of knowledge about web programming is all that is needed to take advantage of it. Rapidweaver makes it easy to include CSS and Javascript in its site-wide and page-specific code areas, and it’s easy to use classes and call functions in most text stacks (not just HTML stacks).

It’s also not that difficult to build a stack around this. Will’s free ‘Builder’ stack makes this really easy, producing a kind of pseudo-stack, but Joe’s great video introduction to stack-building shows how to make an actual stack around a Javascript function. Gary (Doobox) had a useful little Mac app called ‘Stack Developer’ which made the basic setup of a stack really easy — I don’t know if he has any plans to re-release this. The point is, it’s no longer necessary to have the skills and knowledge to write code in order to do this kind of stuff — we can make use of Chat GPT’s ability to generate the code. And messing around with this stuff, if you have the time, is a great way to acquire more knowledge and ability.

A future web-building application will no doubt build this kind of ability in. But what we have right now is already pretty incredible — a giant leap beyond where we were last April. You just have to ask the right questions.

3 Likes

I’ve been chatting to a lot of clients about offering them an AI generate blog service. Lots of interest. So i figured I’d write a blog post about it. Then I though, why not get AI to write it? So I did.

The instruction…

write a blog post about a new service I offer where I will generate a blog post using AI, personalise it, create and add suitable graphics and then publish it to a website.

Reply…

Do you want to write high-quality blog posts but don’t have the time or expertise? If so, our new AI blog post writing service is perfect for you!

Our AI writer can generate blog posts on any topic, and it can even personalize the content to your brand and target audience. Plus, our AI writer can create and add suitable graphics to your blog posts, so you can publish them with confidence.

How Our AI Blog Post Writing Service Works

Our AI blog post writing service is simple to use. Simply provide us with a few details about your blog post, such as the topic, the target audience, and the desired tone, and our AI writer will do the rest.

Our AI writer will generate a high-quality blog post that is tailored to your specific needs. You can then review the blog post and make any necessary changes before publishing it.

Benefits of Using Our AI Blog Post Writing Service

There are many benefits to using our AI blog post writing service. Here are just a few:

** Save time: Our AI writer can generate a blog post in minutes, so you can save time on your writing tasks.*
** Improve quality: Our AI writer can generate high-quality blog posts that are free of errors.*
** Reach a wider audience: Our AI writer can personalize your blog posts to your target audience, so you can reach more people with your content.*
** Increase engagement: Our AI writer can create engaging blog posts that will keep your readers coming back for more.*

Pricing

Our AI blog post writing service is available at a variety of price points to fit your budget. We offer monthly and annual plans, so you can choose the plan that best suits your needs.

Sign Up Today!

To sign up for our AI blog post writing service, simply visit our website and create an account. We look forward to helping you create high-quality blog posts that will help you achieve your business goals.

Testimonials

“I’m so glad I found this service! It’s saved me so much time and effort, and my blog posts have never looked better.” - John Smith

“This is the best AI blog post writing service I’ve ever used. The blog posts are always high quality and relevant to my audience.” - Jane Doe

“I’m so impressed with the quality of the blog posts I’ve received from this service. They’re always well-written and engaging.” - Susan Jones

It’s not my style of writing at all, and I personally wouldn’t use it, but it’s pretty bloody good.

One more thing. These days we all need various kinds of policies on websites, from privacy to terms of sale, returns policies and membership conditions, fair use and forum etiquette. Chat GPT does a great job of creating these (and its results tend to be more concise and accessible than legal boilerplate examples). It’s a good idea to check these against those on other sites, to see if they cover everything you need, and of course to run them past a lawyer (but running something past a lawyer is a lot cheaper than asking her or him to write it!) But, if you‘ve tried to write these things, you’ll know that it is one of the most tedious — ‘maybe I’ll just leave that till tomorrow’ — chores in building a site. Chat GPT will do it in its hallmark ten seconds.

2 Likes

What if it was built right into a stack?

Generate code, content, whatever …

2 Likes

So I have been using the free version of ChatGPT for a couple of weeks. I have found it is very smart but has no wisdom. It’s like a college graduate with no street sense. It often does not know how topics go together in a real life application. A lot of theory but very little experience. Some things it does well. But mostly it is like holding a “new hire’s” hand on his/her first job.

What have you been using it for? I find it rather good at most things I’ve asked it about: its knowledge of languages and literatures, for instance, is astonishing. And whilst it shouldn’t have any wisdom, I often find its answers contain really quite a lot of wisdom. Of course, that is because it has been trained on materials that include much of the wisdom of our species. For this reason I find myself keeping many of the things it has written for me, because they are worth returning to. I have a similar reaction to Midjourney: while the whole genre of computer rendered imagery, from games design to Pixar animations, had an uncomfortable feeling of shallowness for me, like a film set, I find Midjourney’s productions (across its incredibly wide variety of genres) draw me in, as if everything has intriguing depths to it. I certainly wasn’t expecting this of AI, but I am enjoying it enormously.

And here’s a nice little example.

What is the difference between wisdom and knowledge?

Wisdom and knowledge are related concepts, but they are not interchangeable. Knowledge refers to the information, facts, and understanding that one acquires through learning and experience. It can be acquired through education, training, research, and personal experience.

On the other hand, wisdom is the ability to apply knowledge in a practical and meaningful way. It involves a deep understanding of human nature, the world, and oneself, as well as the ability to make sound judgments and decisions based on that understanding. Wisdom often requires experience, reflection, and insight into the complexities of life.

In short, knowledge is what we learn, while wisdom is how we apply what we learn in the real world. While knowledge is important, wisdom is often considered to be a higher level of understanding that allows one to navigate the complexities of life and make better decisions.

Mostly Christian research projects. It’s been great for “keyword“ research or for “category” research. However it sticks things in that have nothing to do or worse are the exact opposite of traditional Christian beliefs. It also likes to stick a bunch of hot button topics in. Often jamming them in where it just simply doesn’t belong.

I have used it on other things. Same result but not as obviously contradictory, rather it is just out of place.

My point is, it’s obvious to me this is just fancy programming with propaganda mixed in.

It’s still useful, but I have to edit at least 50% or more of every 800 words.

One of the themes Chat GPT constantly reiterates is that it is there as an assistant, and indeed this is what it does best. I find it makes excellent summaries of just the right length, which are information-rich but also accessible and easily readable. But I don’t want it to write for me — that would be taking my voice away from me. Reworking what it comes up with is just perfect.

But there is much more that it can do. I asked it to translate a mediaeval Arabic poem the other day: we cycled through different versions, each of which had its merits. I asked it to parse each word of the original text — to identify what part of speech it is, and what it means. This is a huge boon for translating and learning. You could do the same with Hebrew, Aramaic and Greek: now we don’t need to know the original languages to be able to get to grips with the text. That’s an amazing thing!

Once we had a translation I was happy with, I asked it to translate the original in the same way into… I pulled three languages out at random: Italian, Tagalog and Japanese. 10 seconds. Pasting these into Google Translate to check out how well it had done, the re-translations back into English were flawless.

As for propaganda, what you have to bear in mind is that the data sets that these LLMs are trained on are so vast no human being could possibly ‘game’ them. If it seems like Chat GPT has a point of view — and to my mind, that point of view is a refreshing openness — it is because that point of view is implicit in the collective body of human knowledge. Those who have contributed to that body of knowledge, across the ages and around the world, have tended to be rigorous but open minded in the search for truth, pro-human, and concerned with the welfare of human beings. And those are the biases Chat GPT reflects.

For my part, I’m just immensely happy to have lived to see the advent of this technology. And I’m having a lot of fun with it.

1 Like

Mark, you might get (a lot) better results with writing better prompts. There’s a ton of videos on Youtube on ChatGPT prompting and how to improve the results you get.

In addition to what James wrote: aside wisdom and knowledge there’s data and information. ChatGPT is a language model, it uses data. It doesn’t know anything, it uses that bulk of data and depending on the prompt it get, it spits out a, often, coherent string of sentences. Best example of this is when it stops half way and you prompt “continue”, you could also paste part of your own writing, add a blank line and then write “continue” it will finish your writing in the most, kind of, predictable way. I oversimplify, as what’s happening in the background is very complicated and even the creators don’t know exactly how it does what it does. Ok, so back to the bulk of data: because it’s output is based on the bulk of data, it’s also skewed by what data it has been fed and restricted in it’s replies. There are creative ways around it by the way.

And then there’s this… (untried so far)

Chat GPT, Code me a plain Javascript function which takes as input a text, the width of the text container, and the styles which have been applied to that text, and then adjusts the font size of each line of that text separately so that the line exactly fills the width of the text container.

Here’s an implementation of the function you described:

function adjustTextWidth(text, containerWidth, styles) {
  const canvas = document.createElement('canvas');
  const context = canvas.getContext('2d');
  let fontSize = styles.fontSize || '16px'; // default font size
  let fontFamily = styles.fontFamily || 'Arial'; // default font family
  context.font = `${fontSize} ${fontFamily}`;
  
  const lines = text.split('\n');
  let maxFontSize = parseFloat(fontSize); // maximum font size for any line
  let maxLineWidth = 0; // maximum width of any line
  
  // determine maximum font size and maximum line width
  for (let i = 0; i < lines.length; i++) {
    const lineWidth = context.measureText(lines[i]).width;
    maxLineWidth = Math.max(maxLineWidth, lineWidth);
    
    let lineFontSize = parseFloat(fontSize);
    while (context.measureText(lines[i]).width > containerWidth) {
      lineFontSize -= 0.5;
      context.font = `${lineFontSize}px ${fontFamily}`;
    }
    maxFontSize = Math.max(maxFontSize, lineFontSize);
  }
  
  // set font size for each line
  let result = '';
  for (let i = 0; i < lines.length; i++) {
    let lineFontSize = parseFloat(fontSize);
    while (context.measureText(lines[i]).width > containerWidth) {
      lineFontSize -= 0.5;
      context.font = `${lineFontSize}px ${fontFamily}`;
    }
    result += `<div style="font-size: ${lineFontSize}px">${lines[i]}</div>`;
  }
  
  return result;
}

Here’s an explanation of how the function works:

  1. Create a canvas element and get its 2D context.
  2. Set the font size and font family based on the styles parameter (or use defaults).
  3. Split the text parameter into an array of lines.
  4. Iterate over each line and determine the maximum font size that will fit in the container and the maximum width of any line.
  5. Iterate over each line again and set the font size for that line based on the maximum font size and the container width.
  6. Return the adjusted text as a string with each line wrapped in a div element with the appropriate font size.

Note that this function assumes that the text container has a fixed width and that the text parameter contains line breaks (\n) to separate the text into lines. If you need to adjust for different container widths or text formats, you may need to modify the function accordingly.

2 Likes

On this I disagree, but it is what it is.

And there in is the issue.

While Christianity is of course “pro-human” it is centered around whom we believe to be our Creator and His love for us. There are more writings and “raw data” on this subject than any other subject in all of history. It is just surprising to me how “filtered” it is. If not filtered than injected with non-Christian beliefs.

I will check into this, thanks!

With respect, Mark, I think you may be complaining that it is not gamed enough. When it comes to matters of belief, an LLM has to accurately describe the beliefs of a group of people without being seen to either validate or discredit those beliefs. That is a thankless task when a person is asked to do it, but — to my mind, anyway — Chat GPT does it gently and gracefully. But there are exceptions. A facebook friend who believes in conspiracies was very unhappy that Chat GPT had shown little sympathy with the idea of ‘contrails’, and had pointed out that this was an unscientific and unevidenced belief. When I heard that I closed my eyes and said a silent ‘Thank You!’ to the universe.

There’s quite a discussion about it’s neutrality. When it was first launched some used it (deliberately or for fun) to produce extreme output. It turned out ChatGPT was “biased” towards, let’s call it the general view on the world of about 15 years ago (so think pre black lives matter, lbtqxyz±). That was “corrected”, so now it can write lovely poems about Biden but not about Trump, won’t criticize Bill Gates nor Biden (through a/o role playing you can get around this.). In general it’s quite liberal and not so conservative.

I think it’s because of this you aren’t getting the results you’re looking for with a generic prompt, but I also think that the more specific you are the better the results. I get better results when I keep instructions short and building on top of one another.

I am doing research on how to better prompt. I thank you all for your advice.

FYI: I never expected the computer to understand things of the heart and soul. Having said that, it is just surprising to me how contradictory/tainted the information can be.

1 Like

Hej

I think that it is important to understand that AI language models are not conscious or sentient beings; they are advanced computer programs designed to process and generate text. Their operation consists of several stages:

1 Learning how words relate to each other: This helps them understand the structure and meaning of sentences.

2 Reading lots of text: By trying to guess the next word in a sentence, they learn grammar, facts, and patterns in human thought.

3 Fine-tuning: They are adjusted and refined based on specific examples to help them generate more accurate and relevant responses.

Through these steps, AI language models can understand and respond to human input in a way that appears intelligent and natural. However, it’s important to remember that they are ultimately just computer programs analyzing patterns in data.

1 Like

It’s great fun playing with Chat GPT,

“Could you tell me the prinicples of quantum physics in 250 words in the style of Norman Wisdom?”.

But what happens when AI becomes so popular and eventually the norm (no Norman Wisdom pun intended). Where will it derive its references if virtually everything it is gleaning information from is also AI generated.

It’ll be like digital inter-breeding, it will start to mutate and develop deformaties. Now that will be exciting!

It’s worth noting that nothing we’re seeing so far is AI generated. These are very sophisticated and intelligent responses to human queries, accessing a body of work (whether in images or writing) that humans have produced. It would be an even bigger step for AI systems to start generating their own materials, musing on the meaning of silicon life, or creating images from their own imaginations. Personally, I’m not convinced that will ever be possible.

At this stage, I’m much more interested in how we can use what we have here. For instance, if you fancy enlivening your website with, say, quirky flat-color illustrations, you can get Midjourney to produce some really stunning pieces. Illustrators will be, quite reasonably, upset. But I suspect most of us don’t have the budget to pay an illustrator $2000-3000 for, say, twelve illustrations. And just as RW/Stacks made it possible for us to do many of the things that corporations with big budgets were doing on the web, so AI can dramatically up our game again. Multi-language? Fluent pieces without Google Translate glitches are at our fingertips. In the next 12 months, while I’m waiting for StacksPro (let alone an AI powered website builder) there is a whole bunch of really exciting new stuff I can be doing on my sites. Even generating interesting new blog content has been supercharged.

1 Like