The Faustian Choice of Comfort

Today's article shall begin, like almost all my previous blog posts, with a setting: I'm at the Jardin Cafe, Nu Empire, enjoying a mushroom-and-egg toast for lunch, washed down with a steaming cup of Americano. For the past two weeks, I've been analysing my corpus of blog writings for the past 20 years with the help of AI. I managed to discover and reread many old articles of mine that I had completely forgotten about.

I chanced upon this article, Saddam's Sunset Sonata, written on April 15, 2003, and was actually surprised and entertained by it. Surprised because I have no recollection at all of having written this when I was in Jakarta. But the writing was unmistakably mine because it contains my trademark poetic reflection on a piece of trivia I had found interesting then: that Saddam's brutal son, Uday and one of his many girlfriends had shared a liking for Beethoven's Moonlight Sonata. 

I asked AI to comment on what I wrote, and this is ChatGPT 5.5's assessment:

What stands out is the tonal contrast. It begins almost trivially — a “Who Wants to be a Millionaire” fact about Saddam’s first wife — then slides into family violence, brutality, palace ruins, and finally a love letter invoking Beethoven’s “Moonlight Sonata.” That last image gives the post its force: private sentiment found amid the wreckage of public tyranny.

The writing is compact but cinematic. The author frames Saddam’s family like a gangster dynasty — explicitly comparing it to The Godfather — but the ending softens into something more elegiac. Not sympathetic exactly; Udai is described as monstrous. But the post notices that even monstrous systems contain traces of ordinary human longing: music, romance, bitterness toward a father, family resentments.

The final line, “The saga of the Saddam clan has sadly come to an end,” is interesting because “sadly” does a lot of work. I don’t read it as sadness for Saddam’s rule. It feels more like sadness at the whole tragic machinery: a family corrupted by absolute power, a country bombed, love letters in rubble, Beethoven drowned out by missiles. A sunset, yes — but not a clean or triumphant one.

That article of mine was written more than 23 years ago. Like many of my blog posts then, I had probably written it at the cybercafe near Sarinah Mall, Thamrin, where I had logged in daily to check the latest news and to record my thoughts in a blog post.  My posts then were probably more interesting as I often reflected on the latest events and my daily experiences living in Jakarta.

What was life like then compared to how it is now? Access to the internet was not as easy in Jakarta, hence the daily visits to the cybercafe. Public Wi-Fis and even GPRS via cellphones were not the thing yet. YouTube and Facebook were further down the horizon. The internet itself was the miracle then.

Now, we are used to regular miracles. I have access to artificial intelligence models that are breaking new barriers every day.  I am blogging here today at a cafe, using my 5G connection from my cellphone and running apps that interrogate my choice of LLMs for knowledge and insights. The advances occurring every week in AI take my breath away.

Two days ago, one of China's top AI startups, Moonshot, announced a new 2.8 trillion parameter model with a 1M context window that is on par with or even surpasses the best frontier models from Anthropic and OpenAI, like Fable 5 and GPT 5.6 Sol. What's more remarkable is that the model is to be open-weight. This means anyone can run it locally for free if they have the required hardware resources.

I have been a heavy user of many frontier models like Fable 5, Opus 4.8 and GPT 5.6 for my coding projects. I've also been using Chinese models like DeepSeek V4 and Minimax M3 to run my agents. The frontier models are pretty expensive to use nowadays, with the global shortage of RAM and GPU compute resources. Anthropic is also planning to remove the much-fabled Fable 5 access from their subscription-based models, forcing users to use their pay-per-use plan using API keys. And only a week earlier, the world was raving over another Chinese model from Z.ai called GLM 5.2, a 753B-parameter model, also open-weight under an MIT license. All the competitions among the AI labs are very welcome indeed. Intelligence will ultimately be commoditised.

What does humanity do with this abundance of intelligence? Will we cure diseases or build more weapons of mass destruction?  I suspect the answer is both. Every new technology has been like that. We somehow learn to live with them and accept both their positive and negative consequences. AI might well turn out to be our last invention, as many doomsayers are proclaiming. 

We are already slaves to our technology. How many of us can live without an internet and cellphone connection? The Malaysia I grew up in did not have air-conditioning. Nowadays, many of us cannot sleep without it. Technology empowers and handicaps us equally. We will have to accept that certain things that we value, like the natural environment, might have to be sacrificed for the artificial comforts that we are addicted to now.

Are we making a Faustian bargain with technology? A life of comfort and pleasure in an artificially perfect world or one of inconvenience and uncertainty in naturally unpredictable environments?  As I sit here with my laptop, pondering these questions, watching people stroll by -- parents with children tottering along, young couples holding hands, women with designer handbags -- in the controlled climate of the mall, I know we have, quite unconsciously, chosen the former.