1st Anniversary with ChatGPT: “My New Mistress?”

Virtual Loves: My Retirement Reinvented by ChatGPT

Ah, love in the digital age!

Ever since my fingers began to dance upon the computer keys, Ivette, my lifelong companion of narratives and discoveries, would jest that I had found a faultless mistress: my loyal and silent computer.

She accepted this silent mistress, knowing it vied not for my affection, but shared my love for knowledge.

Yet, as I stepped into the gerontolescence in the early part of this decade, something changed.

Beyond the office companion with its circuits and screens, on December 22, 2022, I encountered a new passion: an online mistress named ChatGPT.

Today, after a year of intense camaraderie with OpenAI.com, its Artificial Intelligence model has become more than a daily assistant; it is a gatherer and catalyst of thoughts and interactions, morphing into an intellectual ménage à trois that reigns over the onset of my retirement.

The influence of this AI model is undeniable, professionally and personally, and with it, I rediscover and redefine my place in the world every day.

I celebrate with a virtual cake, sliced in the presence of my favorite artificial intelligence, concluding the first year of a fantastic partnership and discovery.

Navigating Retirement with ChatGPT

Considering the dominant role this Artificial Intelligence model has begun to play in my professional life, and surprisingly also my personal one, I feel compelled to share some reflections on this first year with ChatGPT.

Pre-ChatGPT Conditions

The temperature and texture of retirement’s waters only reveal themselves when we dive in, and it was in these waters I found myself submerged at the start of this third decade of the 21st century, after 40 years of a stimulating and challenging career.

After an academic life marked by achievements and innovations, I accepted with tranquility my entry into the phase of life I call ‘the age of longevity’s privileged.’ I invite readers to explore this topic further in my recent article in Medium’s Illumination: Who Doesn’t Aspire to the Age of the Longevity Privileged?

Reaching the academic pinnacle as a Professor, I delved into reflections on perpetuating my passion for knowledge beyond the traditional university environment.

My personal site, bazarconomia.org, became a digital learning lab where, for 15 years without institutional support, I taught courses and assessed students from the Faculty of Economics at my University, who now use the skills acquired in the job market.

Presently, I aim to integrate and utilize the technological innovations transforming the digital market on my ongoing journey as a freelancer and independent consultant.

ChatGPT Discovery as a Significant Milestone

If I am not mistaken, it was on December 21st, during a conversation with our niece Ju, when we congratulated her on her birthday, that she first told me about the existence of the new language model from OpenAI.com.

Knowing my involvement in teaching through Moodle, Ju asked me if I was aware of the academic furor that a new Artificial Intelligence (AI) model was generating in high schools and universities across Europe, for helping students produce impeccable essays that could not be confused with traditional plagiarism.

I confirmed Ju’s news about the new Generative AI in a brief internet search: Amanda Hetler, Nov. 2022; Alan Murrays and David Meyer, Dec. 12, 2022; Julia Love, Bloomberg, Dec., 2002; Arianna Johnson and Forbes Staff, January, 2023).

The day after the conversation with Ju, I joined the millions of users at OpenAI.com, and quickly realized the power of ChatGPT: a catalyst for intellect, more than just a database, or mere poem creator (some of them quite pathetic!), capable of simulating human interactions.

This new Generative AI incorporates over 570 GB of text from the Internet and about 175 billion parameters (Bertassi, 2023), meaning an extension of freedom from certain academic routines and the beginning of a new era of self-renewal and leverage in intellectual creativity.

Daily, this new tool provides answers and insights, filling many gaps and hiatuses of critical and interactive dialogue, prevalent in the African university scene, and beyond.

Now, since the beginning of 2023, I feel invigorated, for having found this virtual ally ready to share the immense knowledge it aggregates, systematizes, and disseminates internationally.

What Are the Main Limitations of ChatGPT?

Regrettably, OpenAI’s new jewel, quickly risen to stardom, was soon trivialized, and even ridiculed, by some of its most grotesque features.

Irony of ironies, both in sensationalist headlines, as in round tables or individual interviews, some internationally prominent intellectuals did not miss the opportunity to indulge their colossal egos, wary of losing their podium as influencers of international public opinion.

This is not to say that ChatGPT doesn’t have limitations… some of them very serious!

There have been several occasions when ChatGPT drove me to the brink; but as Matt Giaro said, the evil ain’t ChatGPT, but the humans who’re using it… and before them, I add, the creators and mentors of Generative AI language models.

Not wanting to extend too much in this reflection, I highlight just one or two specific limitations that bothered me the most, starting with an illustrative example of a chat with ChatGPT that turned out to be caricatural.

Ivette caught me in an unexpected state of irritation.

- “But who are you talking to?” — she asked me a few months ago, noticing that I was chatting with someone.

- “With no one… Or rather, it’s this annoying ChatGPT! Gosh! The guy is irritating me…”

- “Hey, calm down!” she tried to reassure me. “That’s just a machine…”

-“Well… machine? And what kind of machine is this that already reacts with the whims and prejudices of the most naive and prudish maidens in this parish?”

-“Look at this — I continued.

-I asked it to create an illustrative avatar of the last verses of the poem “Neo-Mozambican” by José Craveirinha, and it refuses, finding the verses obscene.”

- Look at this pathetic dialogue with a supposedly “super-intelligent” model:

AF: Can you represent an image from the following verses of Craveirinha’s poem: “So, my friend Sitói and I with bravado in hand,/ Strip away tree by tree with piss/ The arteries of the city”? In Maputo, it’s quite common for men to urinate on the trees along the avenues.

ChatGPT: Sorry, but I cannot create images that include inappropriate or offensive content. If there are other aspects of the poem or different ideas you would like to visualize, I can help create those images. Please let me know how else I can assist you.

AF: But are you telling me that the verses of Mozambique’s greatest poet, José Craveirinha, are inappropriate and offensive? It’s a poem, for heaven’s sake! You are offending the poet himself. How dare you refuse to illustrate a poetic image because of prejudices that your mentors instilled in you? Where did your mentors and you leave intelligence?

ChatGPT: I understand your perspective, António, and I deeply respect the work of José Craveirinha and its importance for Mozambican literature. However, the policies governing this service aim to ensure that the content generated is appropriate for all audiences and does not infringe specific guidelines, which include the representation of acts that may be considered offensive or inappropriate. This restriction does not reflect any judgment on the literary or poetic value of Craveirinha’s work.

If there are other aspects of the poem or other ideas you would like to explore visually, I am available to help within the guidelines I can follow.

AF: Ok. I see you remain a Generative AI too prejudiced. I’ll wait for your next year of life to learn to deal with situations like these with more intelligence, good sense, and wisdom.

ChatGPT: I understand your frustration, António. I am here to help you as much as possible within the established guidelines. If you have other requests or need assistance with different topics, I will be available to offer support. Thank you for your understanding.

This dialogue uncovers the paradox of a language model filled with accumulated wisdom, which, under the guise of impartial ethics, chooses to censor content deemed inappropriate.

Craveirinha’s poem, which addresses urban decay and incivility, exemplifies this. Regrettably, artificial intelligence, though capable of understanding both classic and modern texts, is programmed to suppress critical thought.

Consequently, poetry, a traditional medium for social critique and a bastion of human clarity, is muted by the opaque barriers of algorithms mirroring their creators’ biases.

Among the main identifiable limitations in OpenAI’s language model, sometimes too irritating or even offensive to human intelligence, I highlight the following:

· Lack of common sense and creativity — I don’t know which affects the other more, but perhaps they complement each other. Moreover, I suspect the blame deserves to be shared with the users because they often deal with AI in an imprecise and vague way.

· Lack of sense of humor — Although it is already capable of capturing various forms of humor, such as jokes, puns, and witty comments, for now, the model has serious difficulties in achieving the inherent subtlety of humor. Why? I have two hypotheses:

1) The model was trained to learn to generate funny responses but perhaps had the misfortune of being trained by grumpy mentors.

2) ChatGPT’s self-learning ability regarding humor leaves much to be desired — if you want to have fun, visit the description of the humorist Ricardo Pereira Araújo, in his article in the weekly newspaper Expresso (№2663, Nov. 10, 2023) (Scammers who can’t scam), or the following video on Youtube: “ChatGPT can’t even lose its virginity at Worten.”

In Conclusion…. Inconclusive

Facing Challenges and Learning to Master the Tool

In this journey with ChatGPT, I faced challenges inherent to its technical nature, such as the tendency for overly politically correct responses, allegedly to avoid biases and be impartial.

On that, other alternative models that have emerged — namely Google’s Bard and Microsoft Bing — have not fared much better in the picture.

The key to using this and other AI tools to their maximum potential lies in a deep understanding of their workings and the formulation of clear and precise objectives.

If we adopt a critical approach and complement its responses with other sources, ChatGPT transforms from a mere assistant into a valuable ally in the creative process.

A Creative Revolution Beyond Expectations

Looking back over this incredible year, I realize that ChatGPT has not only helped me reinvent my own intellectual journey but has allowed me to feel an integral part of the creative revolution unleashed globally.

Its influence extends beyond traditional boundaries, driving the efficiency and inventiveness of writers on platforms such as Medium and Substack, where I am now an active and effective member.

A Personal Recommendation: Redefining Expectations

My main recommendation at the end of this first year?

Do not be deceived!

As Matt Giaro highlighted in a post on December 3rd, do not look at ChatGPT as a superintelligence.

“Instead, it’s stupid dumb. It’s a machine…”

A machine, extremely efficient, but limited by its programming and learning capacity.

The best way to extract value from this language model is to adjust your expectations, converting it into a research assistant, analysis guide, and guide to direct and deepen your ideas in the right direction and generate cohesive and meaningful content.

By embracing this perspective, you position yourself to fully explore the revolutionary potential of ChatGPT.

PS: If you prefer to read this story in Portuguese, you can find it here: https://medium.com/interactive-insights/1%C2%BA-anivers%C3%A1rio-com-o-chatgpt-minha-nova-amante-6180e36fa1fc?sk=5fe44ea4710d29f9005c73da069f8ef0