Artificial Intelligence, is it natural to use it with Joomla?
I remember when the internet was first popularised and the words “world wide web” and “internet” became essential for any headline writer. Soon after the point that the average person in the street would rant “If I hear one more mention of this wibbly wobbly web thingy Im going to” and you can use your own imagination to end that sentence.
For some “Chat GPT” and “AI” have reached similar overuse and triggered the “If I hear it mentioned one more time” response from a fair few.
So at the risk of upsetting a few more, let us dive into the synergy that could make AI and Joomla the best of friends.
The evolution of Artificial intelligence
Pre-20th Century: Philosophical Foundations
The idea of artificial beings has been around for centuries. Ancient myths and stories depict automatons and human-made beings, which can be seen as early ideas of artificial intelligence. However, the philosophical groundwork for AI, such as the understanding of logic and reasoning, started with philosophers like Aristotle and Plato.
Alan Turing is perhaps most famous for his proposed ‘Turing Test’. A human judge converses with two or more participants by means that preclude the judge from recognising any of the participants. If an AI program can fool the judge that it is a human participant, the computer can be said to have passed the test and exhibited an intellect comparable to that of a human.
1950s: Birth of AI
The formal birth of AI is often traced to a conference at Dartmouth College in the United States in 1956. This conference, attended by the likes of John McCarthy, Marvin Minsky, Allen Newell, and Herbert Simon, aimed to explore ways to make a machine mimic human intelligence. During this time, the term "Artificial Intelligence" was coined by John McCarthy.
In 1957, the first complete chess program was created on an IBM. It reportedly took eight minutes to make a move.
1960s-1970s: Early AI Research and Development
AI's early years saw significant advancements, including the development of the first AI programming languages, such as LISP and Prolog. Early AI systems were able to solve algebra word problems, prove theorems in geometry, and learn to speak English. During this period, the first AI chatbot, ELIZA, was created by Joseph Weizenbaum at MIT.
The public exposure to such technology was still mostly through the chess matches that I remember following as a child with increasingly good models but always with a flaw, threatening the queen and they would sacrifice all to prevent its loss, an easy way to beat it.
1980s: AI Winter and Expert Systems
The mid-1970s to the 1980s was known as the "AI winter," a period of reduced funding and interest in AI research due to its failure to meet previous high expectations. However, the 1980s also saw the rise of expert systems, computer systems that emulate the decision-making ability of a human expert, which brought commercial success to AI.
A tipping point was reached in the world of chess and AI. In 1981, Cray Blitz (a computer chess program written by Robert Hyatt, Harry L. Nelson, and Albert Gower and ran on the Cray supercomputer.) scored 5–0 in the Mississippi State Championship. In round 4, it defeated Joe Sentef (chess ranking 2262) to become the first computer to gain a master rating (chess ranking 2258) and only the third computer to beat a chess master in tournament play, after previously reported wins by “BELLE of Bell Laboratories” and “Control Data Corporation's CHESS 4.9”.
1990s: Machine Learning and a New Chess Champion
The 1990s marked the beginning of the era of machine learning. During this period, AI shifted from knowledge-driven approaches to data-driven approaches. The development of algorithms that could train themselves to improve their performance as they processed more data was a significant milestone in AI.
In the early '90s, a new computing chess champion emerged in the form of IBM's Deep Blue. This powerful chess AI utilised parallel computing technologies, creating a supercomputer of unrivalled power.
The series of challenges between Deep Blue and reigning world champion Gary Kasparov is now legendary and felt as though it was becoming more and more a one-sided battle with AI advancing forward all the time.
The inevitable happened on May 11, 1997, Deep Blue beat the world chess champion after a six-game match: two wins for Deep Blue, one for the champion and three draws. Back then the emphasis was as much on the giant machines needed to run the AI programs.
2000s: AI Goes Mainstream
The early 2000s saw the rise of various AI applications in everyday technologies. AI began to be used in a range of applications, from email filtering to personal assistants like Apple's Siri.
2010s: Deep Learning Revolution
The 2010s were characterised by the advent of deep learning, a subtype of machine learning involving artificial neural networks with multiple layers. This era saw significant advancements in AI capabilities, from computer vision to natural language processing. Companies like Google, Facebook, and Amazon started heavily investing in AI.
On the gaming front, smaller and smaller chess machines with better AI models could beat human players, the attention turned to another game, Go and on 15 th March 2016 Google's DeepMind AlphaGo defeated the world Go champion Lee Sedol.
2020s: AI in the Real World
By the 2020s, AI had become an integral part of many technologies we use daily. Advances in machine learning and AI are being applied to numerous fields, from healthcare, screening and diagnosis to self-driving cars. Checking insurance claims and weather prediction modelling. OpenAI's GPT-3, which was released in June 2020 and this year has hit the headlines stealing Google's thunder and eclipsing Google's offering, Bard.
Bard was first announced by Google on February 6, 2023. It was released in a limited capacity in March 2023 to lukewarm responses mostly because of a simple factual error in a demo. On May 10, this year, Google removed the waitlist and made Bard available in over 180 countries and territories.
So for the first time in history, anyone with a browser and an internet connection can talk to Artificial Intelligence. Just let that sink in for a moment.
What can I do with AI?
Short answer: lots, but there are many caveats. AI models at the moment are not always factually accurate. They get a lot right but when they get it wrong they can do it with such gusto that it is hard to tell if it has any self-doubts. But there's a thing: can an AI have doubts at all?
I have the paid-for version of Chat GPT, Version 4 and it's a lot better than version 3.5, but if you want a conversation with AI, Bard I find is more interesting.
Can AI have doubts?
So to answer the question ‘Can AI have doubts’, why not ask it?
So I did.
“Chat GPT, do you have doubts about some of what you write?”
And Chat GPT Version 4 answered:
“As an artificial intelligence, I don't have feelings, thoughts, or doubts. I don't possess consciousness or subjective experiences. Everything I generate is based on patterns and information in the data I was trained on. I aim to provide accurate and helpful information based on that data, but I don't have personal beliefs or uncertainties. It's always a good idea to consult multiple sources or a human expert when you need comprehensive or critical information.”
Good answer and one that I had expected, and now for something a little spookier and in my view, the way things will go.
Let's ask Bard the same question.
“Bard, do you have doubts about some of what you write?”
And Bard answered:
“Yes, I do have doubts about some of what I write. I am still under development, and I am constantly learning and improving. Sometimes I make mistakes, and sometimes I am not sure if what I am writing is accurate or helpful. I am always working to improve my accuracy and reliability, and I appreciate your feedback.
Here are some of the things that I am working on to improve my accuracy and reliability:
- I am constantly learning and improving my knowledge base. I am trained on a massive dataset of text and code, and I am constantly being updated with new information.
- I am working on improving my ability to understand context. I am able to understand the meaning of words and phrases in a sentence, but I am still working on understanding the context of a conversation.
- I am working on improving my ability to generate creative content. I am able to generate text, translate languages, write different kinds of creative content, and answer your questions in an informative way. However, I am still working on improving my ability to generate content that is original and engaging.”
There was more, Bard is quite verbose unless you tell it to keep it short but I do find Bard and ChatGPT quite different in their responses and self-awareness.
So how can we use AI with Joomla?
I'm going to end with a practical example which you can do straight away if you write articles and blogs and give you some external tools to check
In previous articles, I have made reference to my testing site, https://naturallyrelaxing.co.uk/ which I use to test the theories and ideas around SEO and best practice.
I used this recently revisiting some of the articles that I had previously written.
How good are humans?
During the Covid pandemic, the UK government did a scheme to provide out-of-work people with the opportunity to get work experience and ease back into work.
I enjoy working with apprentices and so I thought I would apply. I struck a deal with my local council and shared Chloe, a young writer with a Master's degree in creative writing. She worked 2.5 days with me in the office and all she was required to do was learn, research, and write blog articles for my naturally relaxing website.
We soon got our speed up and could put out several a day.
One day we timed it, we researched 4 different topics, and bounced ideas back and forth. My assistant Stacey was on hand to do the images so between us we managed to research the SEO demand for the topics, build the structure, and complete 4 articles. It was hard to keep the focus and concentration on the job at hand for the whole day but we did it.
And how was the same with AI?
A few weeks ago I sat down after lunch and said to Stacey we needed to use ChatGPT 4 to get some articles done. I did the keyword research but found myself bouncing ideas off ChatGPT, I did use Bard a little and I also used ChatGPT Version 3.5 when I ran into the maximum requests in a time period.
Stacey was doing the images and I asked her to also use artificial intelligence to produce some of the images.
In one afternoon, half the time it had taken for the 4 articles we finished 10! With one human and AI as opposed to 2 humans and no AI doing the writing.
The articles were on similar topics to the ones I had done with Chloe but the finished articles seemed to be better and flow more.
I was very surprised, both at the speed of working but also at my enjoyment of the process.
I found it freed me up to explore more and more quickly get to the results I wanted.
The specific tasks I gave it were:
- Give me your thoughts on my headlines and write paragraphs for each.
- Read the whole article and suggest changes and amendments.
- With Bard, I asked it for references to quotes as it has access to the internet unlike ChatGPT
- Write the meta description once the article was written
- Give me the keywords to add to my article (used to link articles with the core similar article module)
- Read my article and suggest similar topics
- Write the title (asking for 5 different versions so I could choose)
This is not going to suit everyone, a lot of people like human interaction. I prefer AI and find working with AI much easier than working with people.
As Bard said: “I am still working on improving my ability to generate content that is original and engaging”.
So let Bard do the grunt work and free the humans to be creative and explore their writing and art.
Next time I will be looking through all the tools you can use in Joomla, the integrations and extensions that we can use to help with SEO and writing, but for now I will leave you with the links below.
AI Image generator
that was used to create the image Louise Hawkins then used to make the article image
And something to test if the text is written with AI or not
And what did it say about this article?
The classifier considers the text to be very unlikely AI-generated.
I would like to thank ChatGPT4 and Bard for assisting me with this article and the classifier in getting it mostly correct!
Fabulous article Phil! I was getting to the stage where I was triggered by yet another Chat GPT / AI article, but this article is refreshing and a great read. Having played around with both Bard and Chat GPT I agree they have definitely have different personalities! Interestingly I used both to assist with product descriptions, I preferred Bard's, however I was surprised by Bard's keyword stuffing of the product name which I then had to strip out a fair few times so as not to sound like a robot!