Will AI Take Over Our Jobs? A Programmer’s Perspective

As a programmer, hobby musician, and aspiring author, I don’t believe that AI will take over all of our jobs in the long run. Sure, there might be a transitional period where the job market changes, and tasks that are repetitive or require minimal creativity could be fully automated by AI. However, AI is just a tool—a powerful one that enables us to reach new heights and achieve greater goals. I firmly believe that while AI may replace some jobs, it will also create new opportunities. Rather than limiting us, AI has the potential to expand our horizons.

My first profound experience with AI was with ChatGPT 3.5. I was astonished by how well the AI understood me, even when my input was imperfect. The fact that it could grasp the essence of my words and provide insightful responses on almost any topic was mind-blowing. To put things in perspective, I come from an era where I was thrilled to read the BASIC code of ELIZA on a C64, trying to comprehend how this rudimentary program could answer my questions. Of course, ELIZA’s limitations became apparent quickly, but back then, it was a fascinating piece of software.

Today’s AI, like ChatGPT, is far more advanced, but it’s not without its boundaries. After several nights spent chatting with GPT, trying to gauge how close we are to developing a truly self-aware AI, I was relieved to discover that we’re still quite a way off. GPT doesn’t “think” when I’m not interacting with it; it simply generates responses based on the input it receives, aiming to sound as “intelligent” as possible. This often makes GPT appear overconfident. You might ask it to explain a complex problem, and the first response is, “Of course, no problem!”—but what follows can sometimes be complete nonsense. In this sense, GPT can come across as a bit “arrogant.”

Like any powerful tool, we humans need to learn how to use AI responsibly. We can do terrible things with it, like creating deepfakes or manipulating politics, but we can also achieve great things, like finding cures for incurable diseases. Ultimately, it’s up to us to decide how we use this technology. The real danger lies not in the AI itself, but in how people might misuse it. In my opinion, humanity’s biggest problem is a lack of empathy. We’ve become a society where “I” often takes precedence over “we.”

So, will AI take our jobs?

Yes, at least initially. In fields where creativity isn’t paramount, AI will likely render some jobs obsolete. Take music, for example. Much of what we hear on the radio today could already be produced by AI. Tools like Suno and Udio demonstrate that anyone, even without musical talent, can create radio-quality music. The lawsuits from major studios against these platforms aren’t surprising—they fear that AI-generated music could disrupt their business model. It’s also possible that these studios might eventually turn to AI themselves, eliminating the need to pay artists.

What will happen to artists?

I believe that the oversaturation of AI-generated music might eventually push people to seek out music that AI can’t create. This could lead to a resurgence of music that speaks from the soul, performed by real musicians who aren’t just puppets. People who can play live, without any aids, who can truly sing, and who genuinely master their instruments.

Which jobs are most at risk?

Right now, I think print media is the most vulnerable. Texts, layouts, and cover designs can already be handled by students using the right tools, which could drive prices down. I don’t think this will affect painting as much—art has always been about having a real artist with a personal narrative and a unique perspective behind it.

As a programmer, I’m not too worried about AI taking my job just yet. The code that AI currently generates is impressive, but it’s still far from producing stable, reliable code for complex projects. Once a program reaches a certain level of complexity, the call for a real programmer will become loud and clear again. Instead, I believe AI will become a powerful assistant, a modern “rubber duck” that helps us brainstorm and debug.

Let’s not forget that while AI will change the job market, it will also create new roles and help level the playing field. With AI, anyone can learn to program or try their hand at writing or art.

When I first discovered specialized GPTs, like the “Creative Writing Coach,” I had an idea: could AI continue a story I wrote back in 2004? This was a challenge I set for GPT, curious to see if it could expand on an existing narrative.

Can GPT Write a Book?

For those who believe that many books are already being created entirely by AI, I have to disagree. In my experience, AI, particularly ChatGPT, isn’t quite capable of writing a book or even reworking an existing one. The primary limitation is the context window, which isn’t large enough to maintain character consistency throughout a novel. You can go part by part through your text, but the context window is about 150-200 words, from my experience.If you have more text to work at once, you might find that GPT is forgetting aspects of your text, or is confused. This effect will increase significantly if you handle Text with more than 450 words. I find this strange, since everywhere its seen that the context windows is 128k tokens, which should be enough to handle thousands of words. However, it could be that it is limited to the output token size of just 4k tokens. Maybe newer version of GPT will be able to handle it. But GPT 4o was not.

The Issue of Consistency

Here is an example that happened to me. Let’s say you introduce a horse named “Silverwind” at the beginning of your book. By the time you reach the end, you might find GPT writing a scene where Silverwind is holding a torch and shouting, “For Fred!” as it charges into battle. This inconsistency is a significant issue; AI often loses track of character traits or narrative details. It’s as if GPT has a very short attention span, struggling to retain information in its long-term memory. This can lead to characters or plot points being mixed up or entirely forgotten as the story progresses.

The Influence of Training Data

Another concern is the potential for AI’s training data to seep into your story, which could result in unintentional similarities to existing works. This could raise issues around originality and even lead to copyright problems. It’s essential to remain vigilant and ensure that the AI’s suggestions align with your narrative vision and do not unintentionally borrow from the vast array of texts it was trained on.

The Bottom Line

So, can GPT write a book? The short answer is no, and the long answer is noooooo. GPT’s limitations in maintaining character consistency, managing long narratives, and avoiding over-reliance on training data make it more of a writing assistant than a standalone author. While GPT can be a useful tool for brainstorming and generating ideas, it’s not yet capable of handling the complexities of novel writing on its own.

Additionally, GPT’s creativity is somewhat limited. Characters often end up being one-dimensional and predictable. A character facing tough choices, someone who fails or falls into despair, won’t be portrayed convincingly without a lot of effort. GPT tends to generate “Disney-style” stories, where the structure is always the same: tension builds up at the beginning, and it ends with something like, “And with new hope, they faced the future—a future filled with new adventures.”

However, GPT’s language abilities can sometimes be surprisingly diverse, though often predictable. A sunset will always be described with “golden rays” bathing the land in “soft light.” In the end, I was somewhat relieved, though a bit disappointed, that GPT couldn’t continue my story.

Can I use GPT as a writing assistant?

Yes. Where GPT truly excels is in analyzing text flow and offering improvement suggestions. It can provide synonyms or alternative phrasing. Sometimes, you might receive 20 suggestions, none of which are useful—but then an idea pops into your own head. In this way, GPT works as a perfect rubber duck, helping you think through your writing challenges. Also it can provide synonyms and alternatives to your text. It might also be able to help you to structure your chapters and story elements. It can even provide a first feedback of your text. You should ask GPT for a critical feedback, though. Of course we all want to hear how great we are, but if you want to improve your text, you should fokus on the “bad”. However, you should not listen to the suggestions GPT makes, they are often worse than what you wrote.

Because of the limitations that GPT has in form of a very small context window, I used GPT to help me write a python script to split up my text, into smaller chunks:

# splitByWordCount.py
def split_text_into_sections(filename, limit=450):
    # Read the entire file
    with open(filename, 'r', encoding='utf-8') as file:
        paragraphs = file.read().split('\n\n')  # Splits the text into paragraphs based on double line breaks

    # Initialize variables for sectioning
    current_section = []
    section_number = 1
    current_word_count = 0
    
    # Estimate the needed width for leading zeros
    zeroFill = 4

    for paragraph in paragraphs:
        paragraph_word_count = len(paragraph.split())
        
        # Check if adding the current paragraph exceeds the limit
        if current_word_count + paragraph_word_count > limit:
            # Save the current section to a new file
            section_text = '\n\n'.join(current_section)
            section_filename = f"{filename.split('.')[0]}-{str(section_number).zfill(zeroFill)}.txt"
            with open(section_filename, 'w', encoding='utf-8') as section_file:
                section_file.write(section_text)
            
            # Reset variables for the next section
            current_section = [paragraph]  # Start the new section with the current paragraph
            current_word_count = paragraph_word_count
            section_number += 1
        else:
            # Add the paragraph to the current section
            current_section.append(paragraph)
            current_word_count += paragraph_word_count
    
    # Save the last section, if any
    if current_section:
        section_text = '\n\n'.join(current_section)
        section_filename = f"{filename.split('.')[0]}-{str(section_number).zfill(zeroFill)}.txt"
        with open(section_filename, 'w', encoding='utf-8') as section_file:
            section_file.write(section_text)

# Call the function with the filename of your text
split_text_into_sections('Meo-Combined.txt')

If you want to split by chapter instead of breaking into sections of word counts, you can use this script (will be identified by the Title, so you need to insert all titles into this script):

# splitByChapters.py
import re

# List of chapters and their titles
chapters = [
    "PROLOG", "FJITAL", "ZIU", "WERKZEUGE DER MAGIE", "WESTERTHORN",
    "AURASIL", "GORWIN", "DER AUFBRUCH", "AREANDIR", "MORA", "DIE BURG",
    "DER ORTHANBERG", "DUNKELSCHATTEN"
]

# Function to create filenames from chapter titles
def create_filename(index, title):
    return f"{index:02d}_" + title.lower().replace(" ", "_") + ".txt"

# Read the input file
with open('Meo-Combined.txt', 'r', encoding='utf-8') as file:
    content = file.read()

# Find all chapter positions
chapter_positions = []
for i, chapter in enumerate(chapters):
    # Regex to match chapter title with leading number and whitespace
    pattern = re.compile(r'\d+\.\s+' + re.escape(chapter), re.IGNORECASE)
    match = pattern.search(content)
    if match:
        chapter_positions.append((match.start(), i + 1, chapter))

# Add the end of the content as the last position
chapter_positions.append((len(content), None, None))

# Extract chapters and write to files
for i in range(len(chapter_positions) - 1):
    start_pos = chapter_positions[i][0]
    end_pos = chapter_positions[i + 1][0]
    chapter_index = chapter_positions[i][1]
    chapter_title = chapter_positions[i][2]
    chapter_content = content[start_pos:end_pos].strip()
    filename = create_filename(chapter_index, chapter_title)
    with open(filename, 'w', encoding='utf-8') as chapter_file:
        chapter_file.write(chapter_content)

print("Chapters have been saved into separated files successfully.")

Now, after running one of the above scripts to split up the text into separated text files you could upload all txt files in a zip to GPT and then go through them one by one with your prompt. Now, you could rework or analyze your text much more in detail. For example, if you want to find out grammar or spelling issues in your text.

Here is an example of what prompts you could use.

It could be helpful to create a map of characters visual attributes, so you can provide a consistent appearance in your book. This could be stored as a helpful table you can look up if you forgot the hair color, eye color, age. etc.

CHARACTER VISUAL APPEARANCE

Please describe the appearance of [character name] as it appears in the text of chapter [number]. Pay attention to the following details:

Hair: length, colour and shape.
Face: eyes (colour, shape), nose, mouth, ears.
Height and build: Is the figure stout, thin or otherwise described?
Gender and age: Is there information on gender and age or age range? Are specific age characteristics such as ‘old’, ‘young’, ‘between 15 and 20 years old’ or ‘child’ mentioned?
Clothing: What type of clothing does the character wear? Are there any particular styles or colours that are highlighted?
Other visual characteristics: Are there any noticeable scars, wrinkles, birthmarks or other distinguishing features?
Use quotes from the book to support the descriptions and paint a complete picture of the character.

A well-structured prompt for character traits could also address the character’s motivations and conflicts, as these are often crucial to the development and understanding of the character in the context of the story. Here is an example of such a prompt that you can use

CHARACTER TRAITS

Please describe the personality traits and behaviours of [character name] as presented in chapter [number]. Consider the following aspects:

Essence: How is the character generally described? Is he cheeky, funny, quiet, clever, gloomy, melancholic or otherwise characterised?
Role in the world: What position or function does the character have within the story? Is he a main or secondary character? Do they have special tasks or roles in society or the group?
Special abilities: Does the character have special abilities, such as magical powers, superhuman strength, the ability to influence or entertain people?
Gestures and facial expressions: How does the character express emotions? Are there specific gestures or facial expressions that are often used?
Motivations: What drives the character? Are there specific goals or desires that influence their actions?
Conflicts: Is the character facing particular challenges or inner conflicts? How do these arise from his interactions with other characters?
Use direct quotes from the book to support the descriptions and develop a deeper understanding of the character.

In addition to the types of relationships, it might also be helpful to look at the dynamics and history of the relationships. How have these relationships developed over the course of history? Are there important events that have shaped these relationships? Here is an example of a comprehensive prompt that takes these aspects into account.

CHARACTER RELATIONSHIPS

Please describe the relationships that [character name] has with other characters in chapter [number]. Include the following aspects:

Family relationships: Who are this character's family members (e.g. brother, sister, mother, child)? How is the relationship to these people described?
Relationship status: Does the character have a boyfriend/girlfriend, spouse or romantic relationship? How is this relationship characterised - is it harmonious, full of conflict or just beginning?
Friendships and enmities: Who are the character's friends, allies or enemies? How close or tense are these relationships?
Description of the nature of the relationship: Is the relationship characterised as deep and close, distant, conflictual, supportive, or otherwise special? Are there comparisons such as ‘like a brother’ or metaphors such as ‘in love’?
Development of the relationship: How did the relationship develop over the course of the story? Are there key moments or turning points in the relationship?
Influence of the relationship on the plot: How does this relationship influence the character's actions and decisions?
Use direct quotes from the book to illustrate the relationships and provide deeper insight into the character's interactions.

To analyse the role and weight of a character in the story, we could create a specific prompt that looks at both the frequency of mentions and the context and importance of those mentions. This approach would help you evaluate the importance of the character across different chapters and determine whether certain characters may need more or less attention in the narrative.

ROLE WEIGHT

Please describe the role and narrative weight of [character name] in the story, especially in chapters [numbers]. Consider the following aspects:

Frequency of mention: how often is the character mentioned in the chapters mentioned? Are there any chapters in which the character is particularly prominent or barely present?
Significant moments: What key moments or turning points in the story are associated with this character? Are these moments considered central to the development of the plot?
Interaction with major events: To what extent is the character involved in the main events or decisions within the story?
Position in the text: Where in the chapter (e.g. at the beginning, in the middle, towards the end) do important moments occur that are connected to the character?
Influence on other characters: How does this character influence the actions and developments of other important characters?
Use direct quotes and descriptions from the book to illustrate the character's importance within the plot and their presence in the narrative.

A prompt that focuses on examining tension, reading flow and text quality should consider specific aspects of your writing style and text structure. Here is an example of such a prompt that can help you identify strengths and weaknesses in your manuscript:

ANALYSE TENSION, FLOW & TEXT QUALITY

Please analyse the text quality and writing style in the chapters [numbers] of my magical fantasy adventure book. Consider the following aspects:

Tension building: How is the tension built in these chapters? Are there specific scenes or passages that are particularly gripping or, in contrast, less gripping?
Reading flow: How smoothly does the text read? Identify passages that are difficult to understand or disrupt the reading flow, such as unusually long sentences or complicated sentence structures.
Choice of words: Is the word choice consistent with the genre of a magical fantasy adventure? Are there any words or phrases that don't fit the context or style of the book?
Grammar and spelling: Are there any grammatical or spelling errors? Please point out any areas that need correction.
Stylistic elements: Are stylistic devices such as metaphors, similes or symbolism used effectively? Are there elements that should be revised?
Use direct quotes from the book to support your points and offer specific suggestions for improvements or changes that could strengthen the text.

You get the idea. Once you collected all the information you can render a summary.

Its a lot of work and you can consider using a script to do this all for you. However you need GPT API access to do so.

Can I create my own Music with AI and put it on Spotify?

In any case, the quality of the radio music is so low that it is quite possible to achieve the quality of what is played there. The prerequisite, however, is a little knowledge of music genres, patience and possibly some money. Here are some examples I have created using AI:

Just in case you dont believe this was made by AI:

AI singing a stupid text

How did I do it?

First you need to create a song text, A good approach is it to first create a text, that is not in a song-format that provides all the information you want to transfer in the song. Here is an example of sining about a happy monday:

Mondays: the undisputed champion of the weekdays! Who doesn’t love them? While most people are still busy cursing their alarms and escaping the clutches of the weekend, you’re already brimming with excitement to re-enter your work world. Monday morning—a symbol of new beginnings. A fresh canvas, ready to be painted with productivity!

Imagine walking into the office: an oasis of possibilities! There they are, your colleagues whom you feel like you haven’t seen in ages—even though it’s only been two days. Their tired faces are like a gallery of human emotions, a true feast for anyone who loves Mondays as much as you do.

You? You’re beaming with happiness and energy. While others still need their coffee to just survive the day, you’re bubbling over with enthusiasm. And then, the crowning moment of your day: leaving the office with a cheerful “Have a great weekend!” while your colleagues wonder what time dimension you might be living in.

The best part? Coming home on Monday evening feeling like you’ve already put in a whole week’s work. The faces of your fellow humans speak volumes, each a silent witness to the trials of Monday. But you? You’re already looking forward to the next Monday, because every Monday is a new chance to conquer the world—or at least the office.

Mondays, the hero among weekdays, misunderstood and often underestimated—but not by you. Who needs Friday when you can have Monday?

A song can be then generated by GPT:

Verse 1:
Wake up with the sun, it’s Monday morn,
Hit that alarm, no time to yawn.
Fresh coffee brews, the world’s alive,
Into the office, it’s time to thrive.

Pre-Chorus:
While they’re all yawning, you’re adorning,
Bright-eyed, charged, a new day dawning.
Every face tells a story so clear,
But your chapter’s different, let’s get in gear!

Chorus:
Happy Monday, let’s break the mold,
With stories untold and courage so bold.
We say “good weekend” when we depart,
With a Monday heart, we’re off the chart!

Verse 2:
Office lights flicker, your spirit’s high,
They sip their lattes, and you just fly.
Charts and emails, you dance through tasks,
Why love Monday? they always ask.

Pre-Chorus:
You see a canvas, they see a chore,
By noon you’re soaring, they just snore.
Your week’s a journey, they’re just starting,
You play the music, their Monday’s smarting.

Chorus:
Happy Monday, let’s break the mold,
With stories untold and courage so bold.
We say “good weekend” when we depart,
With a Monday heart, we’re off the chart!

Bridge:
Others count days till the Friday beat,
But Monday’s rhythm has them all beat.
You’re the maestro of the weekday race,
Turning every Monday into an embrace.

Chorus:
Happy Monday, let’s break the mold,
With stories untold and courage so bold.
We say “good weekend” when we depart,
With a Monday heart, we’re off the chart!

Outro:
So here’s to the start, not just the end,
Every Monday’s a message we send.
Riding high on this weekday wave,
Monday’s the journey that we crave!

As you can see GPT tries to rhyme extremely hard. Also the texts are quite long. If you tell GPT to create shorter texts, that are easy to sing the output is much smaller:

Verse 1:
It’s Monday morning, rise and shine,
Coffee’s brewing, feeling fine.
Walk into work, I’m on a spree,
Everyone’s tired but look at me!

Chorus:
I love Mondays, can’t you see?
While everyone’s groggy, I’m full of glee.
Good morning, world! A brand new start,
I’m walking in with a happy heart.

Verse 2:
Desks and chairs, the lights so bright,
My colleagues frown, but my step is light.
Throwing high-fives as I go,
“Happy weekend!” I’m all aglow.

Chorus:
I love Mondays, can’t you see?
While everyone’s groggy, I’m full of glee.
Good morning, world! A brand new start,
I’m walking in with a happy heart.

Bridge:
Meetings drag on, but I’m not down,
I wear my joy like a crown.
Five more days till they all rest,
But Monday’s here, and it’s the best.

Chorus:
I love Mondays, can’t you see?
While everyone’s groggy, I’m full of glee.
Good morning, world! A brand new start,
I’m walking in with a happy heart.

Outro:
So here’s to Monday, let’s not pretend,
It’s not the end, it’s just the trend.
I’m riding high on this weekday fun,
Monday’s here and we’re not done!

So now lets focus on how to create the music to this text. Lets move to https://suno.com/ – you can create couple songs for free every day at suno. So, just create an account and enjoy your daily free credits.

Switch to create and activate custom to be able to write your own song text. In “Style of Music” you can write you favorite music style like “surfing, song writer, male singer, acoustic guitar, acoustic, bongos, ukulele, percussions” or “nu metal, electric, power, female singer” – depending of what kind of song you want to create. Here is an example for both:

How can we improve the output?

first of all there some structural things you can do. A good source would be to have a look here: https://sunoaiwiki.com/tips/2024-05-04-how-to-use-meta-tags-in-suno-ai-for-song-creation/

A short improvement would be to have a second voice that can act as an echo or background singer. You can add those by round braces.

[Intro]
Oh, Monday, you champion, rise and shine! (Oh yeah)
Ready to paint this week of mine. (Paint it now)

[Verse 1]
Walking into work, it’s a world anew, (A brand new day, oh yeah)
Faces like art, each mood a clue. (See the clues)
From weekend’s grip, I’m free and clear, (So clear, so free)
Eager hearts, Monday’s here! (Monday’s here!)

[Chorus]
Monday, love you so, you’re my start, (You start the week)
Fresh as the morning, brave as my heart. (Oh so brave)
Sing it out now, let’s not delay, (No delays)
Oh, it’s a brand-new day! (A new day, oh yeah)

[Verse 2]
While they sip coffee, I’m in the skies, (Up so high)
Joy in my steps, no disguise. (No disguise, oh yeah)
Cheer ’til the evening, feeling so grand, (Feeling grand)
A week’s worth in the palm of my hand. (In my hand)

[Bridge]
Stretch out the words, feel the beat, (Feel the beat)
Monday’s rhythm in every heartbeat. (Every beat, oh yeah)
Conquer the office, bright and bold, (So bold, so bright)
Who needs Friday? Let’s break the mold! (Break the mold!)

[Chorus]
Monday, love you so, you’re my start, (Start the beat)
Fresh as the morning, brave as my heart. (Brave and fresh)
Sing it out now, let’s not delay, (Let’s sing now, oh yeah)
Oh, it’s a brand-new day! (Brand new day!)

[Outro]
Every week a chance to be, (A chance to be)
More than what the eye can see. (More to see, oh yeah)
Monday, you’re my hero true, (My true hero)
Here’s to the start, cheers to you! (Cheers to you! Ohh-uhhh-aahhh)

I hope you get the point.

The funny part begins when you try things that no one could sing. Here are some examples that I would consider “funny”:


Sometimes SUNO is also creating a song for you. Here I wanted to count the numbers of PI, but instead SUNO created a strange text in a language thats not a language. I then extended the song to something with a deeper meaning:

Here is the german PI song I wanted to create:
https://suno.com/song/e3c0fbb1-a2f7-48ef-b76b-6fbb6393c185

here is also an example of multiple languages, which was really a challenge. Sometimes the text was more read than sung. However, it kinda worked:

How to upload the song to Spotify?

Ok, till this point you could achieve everything for free. However, uploading your song to Spotify is definitely not for free. There are lots of providers out there, that offers different price models for you. I am using DistroKid to distrubute my music Starting at 22,99 $ / Year you can upload all your music to the mayor streaming services. Its possible to pay for some upgrades. So there is a possibility to pay an extra to your uploaded tracks to keep them for ever, even if you stop paying the fee. This could be helpful to leave a mark after you left the world.

Leave a Reply

Your email address will not be published. Required fields are marked *

*

This site uses Akismet to reduce spam. Learn how your comment data is processed.