The Coming Wave – Mustafa Suleyman

There is so much talk on AI and how it will impact businesses. I wanted to up  my skills a bit and also understand the business implications. The Coming Wave by Mustafa Suleman and Michael Bhaskar was a great read. I have broken down my learning and notes from the book chapter by chapter.

Chapter 1: Containment is Not Possible:

The first chapter sets the tone of the book and gives us a glimpse of what’s ahead.  The discovery of fire and stone tools were the first technologies harnessed by the human species. One thing about inventions, no matter what it is, from pottery to photography to phones and planes, it gets cheaper over a period of time and it proliferates far and wide. This is the nature of technology. The author co-founded a company called ‘DeepMind’ (later acquired by Google) along with his Co-Founders Demis Hassabis and Shane Legg.

Chapter 2: Endless Prolifiration:

In the year 1876, German Engineer Nicolaus August Otto in Deutz AG Factory produced the first functional internal combustion engine, the four stroke model. Otto wanted to use the engine for water pumps and factories. However his partners (Gottlieb Daimler and Wilhelm Maybach) saw an opportunity with transport. However another German Engineer Carl Benz created his version of a four stroke engine of the combustion engine and this is seen as the world’s first proper car. However this was very expensive beyond the means of the very richest. By 1893, Benz had sold only 69 vehicles and by 1900 just 1709. Even twenty years later, there were only 35,000 cars on the German roads. The turning point came in 1908 with Henry Ford’s Model T. Most cars then were priced at $2000 and Henry Ford priced it as $850. Today there are 2 billion combustion engines in everything we use from lawn movers to container ships. 1.4 Billion of them are in cars. This is not just the story of engines and cars. This is the story of technology. Understanding history helps us sketch the future. 

At the dawn of the agricultural revolution, the world population was just 2.4 million. During the industrial revolution it approached 1 billion people (a 400 fold increase). In 1840’s came railways, steamships and more. Think about children who were traveling by horse, burning wood for heat in the 19th century and spent their final days traveling by airplane and houses warmed by heaters (by splitting of the atom)

When Gutenberg invented the printing press in 1440, there was only one press. However fifty years later, there were 1000s of printing presses. The same holds true for electricity, phones and any other invention that changed the world. 

Chapter 3: The Containment Problem:

Can technology be contained? Is it possible to contain technology? The nature of technology is that it will spread no matter what. It is also capable of going wrong often against its original purpose. The makers quickly lose control over their intended use. There are some exceptions. 

In the Ottoman empire , they tried to ban printing technology when the printing press spread in the 15th century. The rulers (Sultan) were opposed to it as they were against the unregulated mass production of knowledge and culture. Istanbul got its sanctioned printing press only in 1727 three centuries after the founding. Nuclear technology was an exception. The nuclear technology being contained is no accident. It is because of the non proliferation policy.  In 1968 nations agreed never to produce nuclear weapons. They were never going to spread like transistors. Building a nuclear weapon is also a very expensive and complicated task a state can embark on.

What about gene editing? Do we want to edit our genomes so our children can have better immunity against certain diseases, more intelligence and a potential to live longer? The author writes that the challenge of technology today is about containing its unleashed power ensuring it continues to serve us and our planet. 

Chapter 4: The technology of intelligence:

The general in AGI refers to the technology’s intended broad scope. AI agents could discover new knowledge which humans might not have thought of.  In March 2016 AlphaGo was pitted against Lee Sedol (a virtuoso world champion) . AlphaGo beat Sedol 4-1.  Information is the core property of the Universe and it could be encoded in a binary format. Foundational breakthroughs in synthetic biology have enabled us to sequence, modify and now print DNA.  We are gaining mastery of atoms , bits and genes. 

Alexnet was designed to classify images.  This was a breakthrough moment for AI. Even at DeepMind, they leveraged on Alexnet’s learnings. DeepLearning uses neural networks loosely modeled after the human brain. In simple terms, these systems learn when their networks are trained on large amounts of data. For Alexnet, the training data consisted of images. It was built by the legendary researcher Geoffrey Hinton (University of Toronto) and his two students Alex Krixshevsky and Ilya Sutskever.  The paper by Hinton and his colleagues became one of the most frequently cited works in the history of AI Research. 

Computer Vision is now everywhere. Driverless cars (classify 2.5 billion pixels per second) which lets SUV navigate through busy streets, amazon check-out less supermarkets and more. This is also the basis of Tesla cars pushing them through autonomy.  Geoffrey Hinton and his colleagues were hired by Google. 

Auto Complete and Large Language Models (LLMs): 

LLM’s take advantage of the fact that language data comes in a sequential order. The challenge lies in designing the algorithm that knows where to look for signals in a given sentence? It autocompletes what comes next. These systems are called transformers. chatGPT was introduced in November 2002 (GPT stands for Generative Pre-Trained Transformer). 

In 1996 thirty six million people used the internet. This year it is over 5 billion people. The author believes that over the next few years AI will become ubiquitous as the internet itself (available to everyone).  The LLM revolution is caused by the fact that very large models can be trained directly from large messy data. Some of these large language models are open source.  In 2022 Microsoft and OpenAI introduced Co-Pilot which makes engineers very efficient with coding tasks. The author writes that by summer of 2020 he was convinced that the future of computing is conversational. Google’s Large Language Model (LaMDA – Language Model for Dialogue Applications).

Alan Turing Test: In a paper published in 1950, the computer scientist Alan Turing suggested a test- whether AI could display human-like conversational abilities that it would be difficult to predict  For example – Go make $1 million on Amazon with a $100,000 investment. The AI may look for trends that are selling, list the products for sale on Amazon marketplace , drop ship items and convert the $1M to $100,000. Though very optimistic, the author thinks this kind of work is possible with minor human interventions in the next one year and fully autonomous in the next 3-5  years. If we pass this test , the implications for the global economy will be profound. Image recognition is advanced and the ability to to write and work with APIs that banks and manufacturers would demand is work in progress. 

Some of the smartest people on earth are working on these problems.

Chapter 5: The Technology of Life:

In the coming decades biology and engineering will converge more.  The author writes that much of the civilization would not have been possible without selective breeding (producing friendly dogs, dairy cattle, domesticated chicken, what, corn and more).  James Watson and Francis Crick, with critical contributions from Rosalind Franklin (through her X-ray diffraction images) and Maurice Wilkins, proposed the double-helix structure of DNA. This structure explained how genetic information is stored and replicated. The purpose of the Human Genome project was to determine the sequence of the DNA in the human genome, identifying and mapping all human genes, analyzing genetic variations among humans and more.  This began a revolution.  Carlson Curve similar to Moore’s law is the collapse in costs for sequencing the DNA. Companies like 23&Me offer DNA profiling of individuals for a few hundred dollars. The power of bio-tech enables us to edit and write codes. Eventually it may be possible to reconfigure ourselves to be more immune to certain diseases or try anti aging techniques. A world where the average life span is 100 years or more is achievable in the next decades. The author asks, What if we could grow what we wanted locally? 

Proteins are the building blocks of life. Our muscles, hair, blood and 75% of our body’s dry body weight is made of proteins. If we are able to understand proteins, we have made a giant leap in mastering biology.

Chapter 6 – The Wider Wave:

In the 1830s John Deere from Grand Detour, Illinois revolutionized agriculture. (Strong and smooth steel was a great material for plowing through dense, sticky soil).  A lot of food was being produced in the midwest.  The future of agriculture is autonomous tractors and robots. The food we eat comes from the world of Robots (in the West), not so much yet in countries like India. Energy is getting more distributed. 

Chapter 7: Four Features of the Coming Wave:

AI as we know is used widely in military and wars. Ukraine was able to sustain itself in its recent war with Russia because of AI and precision equipment.  We don’t have to hit mass with mass, new technologies create pressure points against dominant powers. They develop fast, and are used for more than one purpose and they also have a great degree of autonomy.  The invention of Printing Press meant that a single press can produce thousands of pamphlets spreading ideas with ease as opposed to writing by hand and spreading one by one.Schengen based DJI ‘s drones are also used by the US military. A single AI program can write text as all of humanity. A single system potentially is capable of controlling an entire society. The author writes that in the Coming Wave, a single point program could alter everything. In 10  years a dollar will buy 100 times the computing we are able to buy today. 

Simulations today speed up the process of vaccine discovery. Cello is a framework for synthetic biology design. In drug discovery AI is bound to have a big impact. If AI is the new electricity, then just like electricity it will power our everyday lives. 

We set high level goals for the machine and the AI will figure out how to get to the outcome desired. Keeping humans in the loop is desirable, but optional.The wave is coming. 

Chapter 8: Unstoppable Incentives:  

It was Russia which first encroached on space with the world’s first artificial satellite Sputnik. Twelve years later it was the US (and not the USSR) which put the first human on the moon. Today the US no longer has a big lead in AI Research. Today Tsinghua publishes more AI research than any other academic institution on the planet. China overtook the US in the number of PhD’s produced in the year 2007. China’s R&D spend was just 12% of the US , however by 2020 it was 90%. China is also home to major tech. Companies: Tencent, Alibaba, DJI, Huawei and ByteDance. Advanced Technology is the sharp weapon of the modern state. 

The author also writes on India and predicts that by 2030, India will be the third largest in the world and by 2050, India’s economy will be worth 30 trillion.  Sharing things openly (research findings and more) has its own incentives. The future is remarkably open source (arXiv – documented on GitHub – being built for citations and research)

First Passenger Rail:

In 1830, the first passenger rail opened between Liverpool and Manchester. It had to cross a lot of hurdles, parliament had to get involved, land acquisitions and more. The railway opening was attended by the then UK Prime Minister (Arthur Wellesley) and Liverpool MP William Huskinson. People failed to appreciate the speed of the incoming train and William Huskinson lost his life. 250 passengers a day was being forecast, one month in, 1200 passengers a day were using the railway line. At their peak, railway stocks accounted for more than two thirds of the stock market value. Science needs to be converted to useful and desirable products for people to use it. Simply put, it needs to make money. 

When Thomas Malthus argued in 1789 that the fast growing population would quickly exhaust the carrying capacity of agriculture and lead to a collapse, what he did not account for was human ingenuity. The energy scholar Caclav Smil calls Ammonia, Cement, Plastics and Steel as the four pillars of modern civilization.Climate change is a real problem. To meet this global challenge, we will have to reengineer agriculture, manufacturing, transport and energy systems from ground-up with new technologies so it is carbon neutral or even carbon negative. 

Chapter 9: The Grand Bargain:

Technology alone cannot solve our problems. However it is blurring the lines between nation states. The world has become much smaller because of technology. It has eroded the stable , sovereign borders of nation states. It is a major factor in the deteriorating health of nation states around the world. Synthetic biology and work here will make healthcare more affordable across the world.

Chapter 10 – Fragility Amplifiers:

The Coming Wave (AI) will democratize access to power. Robots will be available to each one of us (similar to how every one has a cell phone).  In November 2020, Mohsen Fakhrizadeh an Iranian scientist’s assaination is attributed to Israel’s  advanced technology including a remote controlled weapon.  The author also writes on DeepFakes – you will not be able to differentiate between a human and a replica. The author also writes on Indian elections where a leader speaks both in English and Hindi and one of it was generated by AI and one could not tell the difference. Deepfakes have a huge problem and it could impact elections. The 1977 Influenza epidemic / also known as the Russian flu is said to have originated from Northeastern China or nearby Russia. Some of these epidemics could just be caused by accidents in labs. The intentions would be great in the labs. However for a variety of reasons there could be unforeseen consequences.Even with Covid 19- Wuhan lab leak could have been the origins of the pandemic.  A lab leak is a great example of an unintended consequence. Biological labs are subject to safety standards. The most secure ones are known by biosafety level 4 (BSL-4) labs. Writing emails, drafting summaries, translating documents , copyrighting – could all be automated. There is a potential for a fall in wages. Today there is a need for prompt engineers.

Chapter 11 – The Future of Nations:

In this chapter the author writes about the Future of Nations.  From Mongols to the Mughals for more than a thousand years the most powerful forces in Asia was the traditional empire. By 1800 that had changed.It was a private company owned by a small number of shareholders. We are in the era of mega corporations that have more valuations than an entire country (Google, Facebook, Apple and more). Apple was valued more than the UK’s FTSE 100 stock exchange combined. Fortune’s Global 500 are already at 44% of the world GDP. Samsung’s revenue represents 20% of the Korean economy. For Koreans, Samsung is almost like a parallel government. The Nation States will use the tools of the Coming Wave to firm their grip on power. When we open our iPhone, it recognizes our face and lets us in. A team of researchers from the Chinese University of Hong Kong founded Sense Time, one of the world’s largest facial recognition companies. China is the leader in Facial recognition technology today with companies like Megvii and Cloudwalk. In 2019 – the US Government banned Federal agencies from buying telecommunications and surveillance equipment from Chinese providers.

In Lebanon, Hezbollah operates like a state within a state. (sizable notorious military wing). The Coming Wave could create more Hezbollah-like entities in fields like education, medicine and more which rely a lot on the social and financial infrastructures. Imagine a future where small groups where in the failing states or anywhere in the world offer AI empowered services like credit unions, schools, healthcare and more?  What if companies themselves start down a journey of becoming states? Or cities break away and gain more autonomy? We will no longer be bound by historic nation states. 

Chapter 12: The Dilemma:

The author writes about catastrophes. The sixth century plague and the 14th century Black Death killed up to 30% of the world’s population. Some of it are man made (World War 1 and 2). Genghis Khan and the Mongol army across China and Central Asia in the 13th Century killed 10% of the world population. The author writes about a possibility of automated weapons with face recognition which can kill people in a targeted manner in busy downtown streets.  Civilizations collapsing are not an exception, it is a rule. A survey of 60 civilizations suggest that they last about 400 years before falling apart. Even a system of States in the best possible health would struggle. 

Part 4: Through the Wave:

The last part of the book talks about potentially containing the AI wave. 

Chapter 13- Containment must be possible:

The author writes on containment of the AI Wave. He talks about regulation which has made cars, planes and medicines safer. The European Union’s AI Act was first proposed in 2021 and is going through a lengthy process to become law. Complex regulations refined over decades have made our roads and lives safer.  The EU restricts genetically modified organisms in the food supply. However in the US, they are a routine part of the Agri business. China limits the time children under 18 spend on games and social apps to 90 minutes a day during the week and three hours on the weekend. Meta would not share the view that social media is a part of the problem.

Chapter 14: Ten Steps towards containment:

The author writes about funding more initiatives towards AI and Bio Safety. He shares an example of how photo copiers and printers restrict printing or making photo copies of money. He writes that the wave can be slowed in some areas. TSMC’s machinery used to make chips come from a single Dutch supplier ASML, Europe’s most important and influential technology company. Technology Solutions are not just sufficient. The author writes on new tax on automation and autonomous systems – tax on robots? (Note: I am not sure if this is efficient / effective). We can also learn from the airline industry. Every accident is a learning experience and they keep refining. This is not kept proprietary and is shared across all airlines. The author writes that even the President of the United States has limited power to contain the internet. People fight for change. (e.g. Slavery, civil rights, women suffering and more). Change happens when people demand it.  

The author ends the book with the Story of Power Loom. In 1785, Edmund Cartwright invented the power loom, a new mechanized way of weaving. It did not catch on. However with iterations, the technology developed. The power loom that was invented could be operated by a single person producing as much fabric as three and half people. This did cause a lot of distress and many people were not happy with this. This meant that the wages of weavers were halved while the living costs increased.  Textile workers felt that their jobs were being taken away. In 1807, with the support of Ned Ludd, weavers across the English country organized protests. The protest turned violent. This is also called the ‘Luddite reaction’ However, protests led by Ned Ludd started to taper down. England around this time only had a few thousand automatic looms. By 1850, the number became a quarter of a million. The battle was lost. In the long run, the same industrial technologies that caused so much pain have helped make great improvements to our living standards. Even the descendants of Ned Ludd would be very happy with the progress.  

My Take on the book:

Thanks Mustafa Suleman and Michael Bhaskar for writing ‘The Coming Wave’- I not only learned about what is to come, but really liked how you also gave some historical context with inventions (from combustion engines to power looms). As you write in the book, Change will happen when people demand it. Let us ride the coming wave and fix problems that come our way. Thank you.