The dawn of mainstream AI… or getting a computer to write a blog for you

The public launch of ChatGPT and AI powered image software in late 2022 clicked as a new topic to explore in a blog after it made headlines, and I had a chance to test it out. The promise of AI as a world changing force has always been limited to the realm of science fiction. Images of rogue computers or sentient robots taking over the world or manipulating people in various ways has been the theme of many science fiction novels and films over the years. ChatGPT is the first public look at a powerful writing tool that rivals human capabilities for accuracy and even creativity. AI generated images are filtering into social media feeds to the point where it is difficult to determine what is real and what is computer generated.

AI Generated image of Author — Jeffrey Mackie

The most common measure for whether an AI has reached ‘human level’ intelligence is whether it can pass a ‘Turing Test’. The basis of a Turing Test is whether the user of the computer testing the AI system can’t tell if the responses are coming from a computer AI or a human.

ChatGPT or Chat ‘Generative Pre-Trained Transformer’ is a creation of the company OpenAI that generates text responses to virtually any written query. While on the surface this may seem like a basic tool, as computers have generated responses to text since the first text-based computer games. However, the AI generated responses from ChatGPT are well-written, thoughtful and even creative. You can ask it questions, instruct it to draft letters, emails, poems, stories or any type of text you want in any style you can think of, and it responds in seconds with surprisingly high-quality and detailed answers. This blog will show a few simple queries along with the ChatGPT responses, so you can see for yourself.

After interacting with the system and testing it out in my own hypothetical ‘Turing Test’, I would argue that ChatGPT wouldn’t pass off as a human. Not because the answers aren’t plausible or full of errors, but because no human can generate pages of well-written and error-free text on nearly any topic in a matter of seconds. Anyone who has ever interacted with a ‘chatbot help desk’ or tried to navigate the voice command menus of an automated call-centre, will attest to the fact that they can’t get a human on the line fast enough. With ChatGPT, AI technology has leapfrogged from being obnoxiously bad to the point where it is faster and often more accurate than a human providing a response.

Perhaps if the program was slowed down significantly and added random typos and grammatical errors it would do a better job of passing a ‘Turing Test’ and providing a true ‘human’ response. But why would we want to use fallible humans as the gold standard in the first place? The goal of most technologies is to make things better, faster or more efficient, not error prone and emotional.

I suggest trying out Chat GPT for yourself at the link below. Type in your favorite topic or something totally absurd and see what it comes up with.

(Be aware there are a number of phony iPhone/Android apps that look like they are published by ChatGPT/OpenAI and charge a fee or subscription to use. The phony apps are simply interface tools that send the query to the ChatGPT website in the background and present the results in a phone app. The official access to ChatGPT is free on their website (for now) and can be accessed on any web browser, including the one on your phone.)

What better way to test out the capabilities of the AI than to ask ChatGPT to write the blog for me in my own writing style? (the Bold text is my input and the Italic is the response from ChatGPT)

“write me a blog in the style of Jeffrey Mackie about AI writing a blog to see if people can tell the difference”

Hi there! I'm an AI, and I've been asked to write a blog to see if people can tell the difference between my writing and that of a human.

As an AI, I have access to vast amounts of information and can process it at a much faster rate than a human. This means that I can quickly research and analyze topics, and generate written content based on that information. But the real question is: can I write in a way that is indistinguishable from a human?

Well, I'm about to find out. In this blog, I'll be discussing the potential benefits and risks of using AI in various industries. I'll be analyzing data and drawing conclusions based on that data, just as a human might.

So, as you read this blog, think about whether you can tell that it was written by an AI. Do my conclusions and arguments sound logical and well-supported? Do my writing style and tone seem natural and human-like? Or do you suspect that a machine is behind this content?

It's important to note that AI has the potential to revolutionize many industries, from healthcare to finance to transportation. By automating tasks and analyzing data at a faster and more accurate rate than humans, AI can help businesses and organizations to become more efficient and effective.

However, there are also risks associated with the use of AI. One concern is the potential for job loss as AI takes on tasks that were previously done by humans. There is also the risk of bias in AI systems, as they can only be as unbiased as the data they are trained on. It's important to carefully consider these risks and take steps to mitigate them as we continue to develop and use AI.

So, there you have it — a blog written by an AI about the potential benefits and risks of using AI in various industries. Can you tell that it was written by a machine? Let me know in the comments below.

After I hit go, the AI generated the text above in about 3 seconds. The response is unedited or changed in any way other than switching the font to italic to note the difference between my writing and that of the AI. Without the obvious ‘first person AI’ perspective would you have known it was written by a machine?

Did it really ‘read’ the roughly 50 blog posts on my website and tune the writing style to mimic my own voice? It is hard to tell how the text was created or what sources were used, but for a software program to respond on a random topic in a matter of seconds and create a 5 paragraph ‘blog’ in a casual tone that resembles my own writing style in a reasonable approximation, it is hard not to be impressed. And this is just the first glimpse of what this technology can do.

What does this mean for students, professionals, lawyers, writers or journalists? Could any student have a machine write a paper on any topic in seconds? The answer to this one is yes. And more importantly, will a high school teacher or university professor be able to tell the difference when grading the papers? Perhaps the professor could tell some of the time, but what about a hybrid approach where the student has the AI write a first draft then cleans up the text and makes final edits? Is this an ethical approach to writing a term paper? How is something plagiarized if a machine wrote it and nothing was copied? This clearly enters a gray area of what is an acceptable submission as a student’s own work.

One could argue this technology is simply a new tool just like when math students moved from manual pencil calculations and slide-rules to electronic pocket calculators for solving math problems. What was once a revolutionary technology is now a simple feature included on any smartphone. On the other hand, solving a square root or simple geometric function is much different than generating an entire term paper. The math student still needs to understand the fundamental concepts to know what to input to get the right answer, while the AI generated term paper could likely fool many professors as an authentic piece of student work.

I can imagine dozens of uses for this type of technology as it is further refined. An AI could draft responses to emails and then I could quickly look them over myself to control for accuracy before a reply is sent. It might not be there yet, but with further refinement it will certainly achieve this capability in the next few years. Taking it one step further, it will likely develop to the point where an AI program could automatically reply and manage my email inbox entirely without my input required at all. Will email eventually become a world of AI Chatbots sending emails back and forth to each other?

The ChatGPT is not trained to comment on current events and only has data loaded up until 2021, but will it eventually be able to write newspaper articles and comment on current events? How will we know what is real news and what is AI generated? (I suppose we could ask that question about the news media today without the influence from AI systems …)

How creative can the AI get in its responses? Will it be able to write full novels, screenplays, poetry or other creative works?

Let’s test it out — Here is the input requesting a poem about trees:

“write me a poem about the beauty of trees”

Trees, oh trees, so tall and grand, Reaching towards the sky with open hand.

Their branches stretching wide and free, Their leaves a rustling symphony.

They stand so strong and tall and true, A testament to what they can do.

They give us shade and oxygen too, A gift that we cannot overdo.

Their bark rough, their leaves so green, A sight that's always fresh and clean.

Trees, oh trees, a beauty to see, A natural wonder for you and me.

So let us take a moment to stand, And appreciate these giants of the land.

The poem above was also generated by ChatGPT in a few seconds. It’s a bit corny, but admittedly a solid poem that accurately covers the topic and the AI seems to ‘know’ the important attributes of trees. I’m certain I couldn’t have come up with a better poem about trees, and even if I could it definitely would have taken much longer than a few seconds. Will the AI get to the point where it can write an entire novel or screenplay? How about combining an AI generated story along with AI generated images and animation?

AI Generated image of Author — Jeffrey Mackie

The image above was generated by uploading a dozen images of myself into an AI system and it came up with 50 images of me in various styles and scenes. Here I am, apparently starring as an astronaut in the next Hollywood blockbuster. The images in the blog are a small sample of the result generated by the image AI called ‘Wonder’.    

Could an AI come up with an entire plot, dialogue, characters, animations and digital worlds that rival the works of Pixar or the major animation studios? Again, the technology isn’t there yet, but individual components are, and putting them all together is a likely outcome in the next few years.

Perhaps AI generated investment advice will become a threat to my industry. Let's see what it comes up with:

“write me an economic forecast letter for the Canadian economy in 2023”

Dear valued clients,

As we look towards the year 2023, the Canadian economy is showing signs of steady growth and stability.

In the past year, we have seen a rebound in key industries such as manufacturing and tourism, as well as a steady increase in employment and consumer spending. This bodes well for the Canadian economy moving forward.

In terms of specific sectors, we expect to see continued growth in the tech industry, as well as a potential resurgence in the energy sector. The housing market is also expected to remain strong, with modest price increases and low mortgage rates.

However, there are also potential challenges on the horizon. The global economic landscape is uncertain, and any unexpected events or disruptions could impact the Canadian economy. In addition, ongoing trade tensions and the potential for further protectionist measures could pose a threat to Canadian exports.

Overall, while there are risks to consider, we believe that the Canadian economy is well-positioned for growth in 2023. We will continue to closely monitor developments and provide updates as needed.

Thank you for your continued trust in our company. If you have any questions or concerns, please don't hesitate to reach out.

Sincerely, [Your Name]

Would you be able to tell if the text above was written by me or a computer AI? What if I had started with this text as a rough draft and added some specific data points and made a few edits to the structure, style and content? One could argue using a tool like this is a more efficient use of time, allowing me to provide better service and advice in a more efficient way. Others might say it is cheating and not coming up with original content. Where do the ethical boundaries of the use of this technology start and end? This blog seems to raise more questions than answers, but ChatGPT has only been live to the public for a few weeks, so it is still early days and much too soon to tell what the long-term impacts across industries will be. As this technology sees wider adoption it will certainly be a topic of discussion and fascinating to see what others come up with using this new tool.

Welcome to the future.

AI Generated image of Author — Jeffrey Mackie