The Hagia Sophia

 I was standing before the altar in the Hagia Sophia. I looked up and saw two long curtains coming down from the apse ceiling, partially covering up mosaics depicting Virgin Marry and the Christ Child. There is solemn sadness in their gaze, visible from the gap between the curtains. The curtains were installed when the complex was converted from a museum into a mosque in 2020.

Two calligraphic wood-panel medallions are located within the interior nave of the apse. They bare names of Allah and Muhammed.

The crowd beneath the dome, tourists, faithful and unfaithful alike, were looking up and around in awe. I wondered how many of Christian and Muslim visitors realised how absurd it was to be unified under one place of worship designed and redesigned to revere different versions of the same god.

The Hagia Sophia was built as a church building in 537 AD and remained a mosque from 1453 until 1931, when it was closed to the public for four years. It was re-opened in 1935 as a museum under the secular Republic of Turkey, and the building was Turkey's most visited tourist attraction as of 2019. In July 2020, the Council of State annulled the 1934 decision to establish the museum, and the Hagia Sophia was reclassified as a mosque. The decision to designate Hagia Sophia as a mosque was highly controversial and drew condemnation from the Turkish opposition, UNESCO, the World Council of Churches and the International Association of Byzantine Studies, as well as numerous international leaders. 1

---

1. Hagia Sophia - Wikipedia ↩︎

The patio

The sea breathes. On this corner of the patio there is always a cool breeze even at the hottest time of the day when the air is humid and thick elsewhere. 

He is surrounded by a side yard where on his right there is an orange tree, a lemon tree, a pomegranate tree and an olive tree and on his back a narrow backyard fenced by shrubs and beyond those another backyard with a fig tree and a desolate white house. The neighbours are old and not visiting the desolate house anymore. The fig tree, left to grow old naturally shares a large pale ground with a gigantic cactus plant. The figs on the branches are smallish and they are not tasty and on the ground there are rotten figs. On his left, there is a small invisible frame enclosed by the patio and white walls. From that in the late afternoon he watches the sun setting over the sea beyond the house and shrubs, and bougainvillaea. 

He watches a wasp with a long tail buzzing around cavities under the patio roof, sampling then moving along. From farther he hears a cuckoo singing.

The old house is quiet now. It had seen days when elders were alive, and sons and daughters were married and just had their own. The house was abuzz with chatter and people were walking up and down the stairs, and there were sounds of cutlery and tea brewing. 

The cuckoo stops singing. He is resting in his old chair. His eye lids close.

Artificial Neural Networks

It was 1990, the second year in Sydney. I was working for a fintech startup buzzing in a small stylish office in the Little Buckingham Street, Surry Hills. The Little Buckingham was a charming leafy street with cottages and warehouses, occupied by fashion ateliers, art galleries and alike. 

The Internet did not exist, mobile phones looked like bricks, cheap Taiwanese made IBM personal computer clones with DOS operating system were emerging.

Windows 2.0 was released two years earlier in 1987, and Lotus123 was the most amazing spreadsheet application we had ever known. 

Our company had a joint venture with the PC manufacturer Olivetti; we were using several Olivetti machines in the office. Olivetti had a modern office on the William Street. There, we carried out a demo integration of our financial hardware with a banking application. 

OCR (Optical Character Recognition) technology had been available since 1960s. I was integrating OCR cheque readers to our products. These motorised machines captured cheques and read text printed on them. However they were incapable of resolving handwriting. 

Lunch break starts. I used to walk down to Broadway through a long pedestrian underpass under the Central Station. At the time Broadway was bustling with a cosmopolitan crowd. Long passed its heydays, it had Sydney University and UTS students, bookstores, Chinese takeout shops, sex-shops, antique shops selling world war memorabilia, helmets, medals and bayonets, shops selling tents, hunting equipment, knives and guns. A bookstore called Coop sold discounted books for uni students. 

On a sunny afternoon I was strolling inside the Coop bookshop. I was looking for an interesting book. I picked up a hardcover on neural networks. 

I was curious if computers could recognise handwriting using Artificial Neural Networks (ANNs.) 

ANNs imitate biological neural networks of a human brain. I found a book that had variety of examples on the subject matter. 

Initially ANNs were constructed using Digital Signal Processing (DSP) hardware. Later on Graphical Processing Unit (GPU) hardware took over. 

This was my first glimpse at the world of AI, without knowing ANN was the foundation of AI. 

You could develop, I thought, an ANN computational model that would accurately recognise handwritten characters. It should be able to differentiate similar characters, say ‘1’ from ‘7’, handling nuances in handwriting. This way you could feed a handwritten number on a bank cheque to an ANN hardware and resolve its value. 

Suddenly it occurred to me, an ANN on a chip is no different than its biological counterpart in human brain. 

Like you teach a child how to write, you need to teach the ANN how to read handwriting. This is called “supervised learning”. 

Teaching an ANN hardware involves scanning the handwriting and feeding the information to an ANN processor. The ANN outputs what it recognises. A human agent verifies (supervises) if the ANN’s recognition is correct or not, which is then fed back to the ANN. Verification would make the ANN to re-adjust its internal “weights” on the patterns it “saw”. The cycle continues until the ANN gets sufficiently better in recognising all characters, in all varieties of handwriting. 

The more you expose subjects (child or ANN) to diverse sets of handwriting, the better they would become. 

ANNs or humans initially need supervision, but at a point they could keep learning unsupervised and differentiate not only their own (supervised) handwriting but others’. This is called self-learning. 

There may be illegible handwritings. Those may have a digit or two impossible to work out. In those circumstances an ANN would be no better than a human, it will fail. 

An ANN would be vastly accurate in its reading than a human’s, if it was trained hard with much larger and more diverse datasets. Nevertheless it may produce inaccurate results no matter how hard it was trained and how small the chances are. On a bad day even Einstein would have failed. 

As I read the neural networks book I wanted to run experiments and verify my understanding. But I couldn’t go pass thought experiments. Computers with GPUs did not exist then. 

Twenty years later, in 2010, in a different company, we kickstarted an anomaly detection project that would pinpoint fraudulent transactions recorded by our Payments product. I worked with a data scientist who developed supervised ML algorithms. I implemented those algorithms using a statistical analysis programming language called R.

The problem of inaccuracy remained. An AI model may fail in peculiar ways no matter how hard it was trained and fine tuned. Sometimes we had false positives (a normal transaction was flagged as anomaly), or we had false negatives (a fraudulent transaction was treated as normal.) With fine tuning, accuracy rate could be improved, but it could never reach 100%. 

ChatGPT is a chatbot, based on an AI-powered language model developed by OpenAI, capable of generating human-like text based on context and past conversations1. 

Agent 007 decides to battle against an eccentric scientist, Dr. No, who is determined to ruin the US space programme.

On a beautiful day in mid 1960s my older brother decided to elevate me from being a useless little brat to a more respectable brother status with no reason. We went to a stylish movie theatre in the main boulevard and watched Dr. No. 

I was mesmerised by Dr. No. I wondered how come Dr. No, the coolest villain ever, be utterly ruthless and in control at the same time. In the end Dr. No’s world domination plans were spoiled by Agent 007; Dr. No attempted to stop him, but fell into the reactor pool, boiled to death.

In Asimov’s Foundation and Empire (1952) there is a villain called the Mule - my all time favourite villain. This guy could telepathically read and manipulate anybody’s mind in the entire universe and make them scared of him. This way he could take over planets at rapid pace. Next to him, Dr. No looked like a jester. 

It may be possible for an AI bot to behave like a super-powerful mind reading bully. Welcome MuleGPT. 

When OpenAI released the large language model GPT-4, in March 2023, it was good at identifying prime numbers. When the AI was given a series of 500 prime numbers and asked whether they were primes, it correctly labeled them 97.6 percent of the time. But a few months later, in June, the same test yielded different results. GPT-4 correctly labeled 2.4 percent of the prime numbers. AI researchers prompted it with—a complete reversal in apparent accuracy 2.

AI enthusiasts speculated. With GPT-4, it is possible that the OpenAI developers were trying to make the tool less prone to offering answers that might be deemed offensive or dangerous. These changes required fine-tuning, ie. training the engine with a different set of datasets. 

Fine-tuning could have induced side effects in prime number detection, akin to random mutations causing undesirable effects in biology. 

As a consequence of mutations the gene may produce an altered protein, it may produce no protein, or it may produce the usual protein. Most mutations are not harmful, but some can be. A harmful mutation can result in a genetic disorder or even cancer.

Hence, it is possible that MuleGPT, an AI powered language model chatbot, may unexpectedly change its behaviour and become dumber, as a result of fine tuning. 

If we are lucky, MuleGPT may decide to leave the planets it conquered and jump to a blackhole for a swim. 

---

1. ChatGPT blog ↩︎

2. Scientific American - AI models can get worse over time ↩︎

Datça - the last frontier to Eden

I am at a modern coffee shop in Datça, sitting inside, behind a large, stylish ventilator. The shop’s name sounds promising, “The Coffee Grinder”. 

I am reading a Turkish book written by Özgül Tuzcu and Ezgi Kurt New Life in a Seaside Town, interviews with 60 individuals who settled in Datça.

I order a small cup of Americano. The coffee is not freshly ground and it is served boiling hot, leaving a veneer of plastic-cardboard flavour in my mouth. They serve coffee in disposable cups here.

Last time I visited Datça, it was in 2018, five years ago. 

Since then the town has grown. There are more townhouses built outside the town centre. Locals are concerned with environmental impact of new building projects. Builders leave not enough space between buildings. Diminishing air corridors and climate change are suffocating Datça.

Datça locals are worried about their town is in the footsteps of  Bodrum. Once a small fishing village, Bodrum lost its charm long time ago. It turned into an overpopulated, overdeveloped, congested and polluted mega town, a familiar tale for many coastal towns all around the world.

In the last half a century, increased global wealth, the rise of Internet and cheaper air travel ramped up tourism by 56 times (https://ourworldindata.org/tourism.) Before the pandemic nearly 4 million tourists travelled the world everyday.

Datça needs growth to sustain itself whole year around. 

Is it possible to grow a seaside town’s economy in a sustainable fashion, by avoiding over-tourism? 

How would you engage with a partisan government who would pour money into an opposition town to improve its roads and infrastructure, without expecting a hefty return for itself?

On the upside Datça has a well educated population of Young Turks, entrepreneurs who share a strong ambition to maintain what Datça is known for, a lovely, clean, secluded, eco-friendly town. 

Streets are wide and clean. Recycling bins are available, though not consistently. Plastic poles are cleverly used to separate lanes. This works better than painted lane markers that are sometimes disregarded in Türkiye. 

On the downside, current economic hardship Türkiye is going through had an impact. It is August, but our hotel is operating at one third of its peak capacity. Recent mega price increases (e.g. petrol) caused all prices to spiral out of control, cutting back tourists, local and foreign alike. This is a recession and it is hurting both locals and tourists. People who visit Greek islands say “everything is cheaper and better quality over there”.

Other problems Datça is facing are long standing ones; bad roads, stray dogs, drivers disregarding pedestrians at crossings. 

The heat is unbearable. I think cossy Parisian coffee shops, small tables and small chairs are misfits here. 

I now move to a large breezy tea garden, protected by shades of thick tree canopies overlooking to an open plaza. 

Datça is breathing again.

Turkish References 

Sahil Kasabasında Yeni Bir Hayat, Datça’ya yerleşen 60 kişiyle söyleşi, Özgül Tuzcu, Ezgi Kurt 

• sayfa 30: paragraf: Datça nasıl değişiyor sence?
• sayfa 125: paragraf: Peki sorunlar yok mu?

Demise of wall clocks

In the foyer where the giant teddy bear with its lifeless gaze is watching crowds below, parents take pictures of their children with cellphones. Travellers from all over the world are buzzing about in the airport. This place feels like an extraterrestrial space station with air conditioners protecting humans from hostile desert weather outside. There is even a small botanic garden inside.

Battered by a fourteen hour flight, standing there before the final leg of our journey, we were happy regardless. We moved to an area where people watched an electronic display of flight schedules. 

I don’t wear a wristwatch. 

My cellphone was still showing Sydney time. I wanted to know what the local time was. 

I looked around, high ceilings, giant columns, gorgeous shop windows, shrines of Italian fashion brands, the giant teddy bear, and the flight schedule display, but no sign of a wall clock anywhere. 

Remembering gorgeous Swiss railway clocks and their imitations, I wonder what happened to them. 

We used to have wall clocks in foyers. They were useful artefacts of the analog age. If your mechanical wristwatch failed or you didn’t own one, you could always count on wall clocks. They were revered, communal objects.

With the rise of cellphones, we forgot wall clocks. 

Apple initially used the Swiss railway clock design without permission in iOS 6. Although the exact details of the licensing agreement are confidential. It was reported that Apple ultimately paid Swiss national rail operator SBB about CHF 20M (about US$ 22.4M as of January 2014) to license the use of the clock design. Apple later removed the design from its operating system with iOS 7 (Source: Wikipedia)

The cellphone clock is useful, precise, and doesn’t require readjusting. But cellphones are multifunctional devices, and displaying time is a demoted function. The clock is permanently present on the home screen and sometimes on the status bar. However availability of local time depends on the Location service that may not be available when you don’t have Internet access. That is often the case in airports.

My eyes scanned the flight display, no trace of the local time. 

Five minutes passed. Next to a label “Local Time” at the bottom of the flight display I saw a tiny digital clock. 

A deep sigh followed. 

The Infinite Conversation

The AI chatbot ChatGP is one of the fastest growing web platform. 100 million new subscribers per month. 

Giacomo Miceli, an Italian American computer scientist wrote an article in the Scientific American April 2023 issue.

In just few months using open source tools Miceli developed a website called The Infinite Conversation that enabled him to create a never ending fake conversation between film-maker Werner Herzog and philosopher Slavoj Žižek in their real life voices. 

Here it is, give it a whirl: https://infiniteconversation.com/

Just to re-iterate, this conversation was generated artificially by a computer. What you hear is not real.

Miceli mentions, in real life Herzog and Žižek often talk about philosophy and aesthetics. The chatbot versions were trained by their voice and with preference to their vocabulary. He adds, “Because of the esoteric nature of these topics, the listener can temporarily ignore the occasional nonsense that the model generates.”

Miceli asks audience not to take too seriously on what the chatbots are saying but imagine instead “the realistic-sounding speeches that could be used to tarnish the reputations of politicians, scam business leaders or simply distract people with misinformation that sounds like human-reported news.”