Book review: Teaching Machines how To Cry by Dr. Paula Hidalgo-Sanchis

 

Review 

Teaching machines how to cry by Paula Hidalgo-Sanchis was recommended to me and I found it different. The shell on the book cover made me wonder about the book itself or the untold story. A very unsophisticated cover, but once you follow the story, the meaning becomes clear behind the simplicity. Life is very fragile, the outer casing only shell deep, and if you put too much pressure on it, it will break. The question was, how does this apply to the characters within the book? 

The book is the debut novel from the author, a highly esteemed person with a biography to impress. The relationship between Alba and M is very significant. Hence the name of the book, Teaching machines how to cry. It all began when Alba became ill, and an AI prototype became part of her everyday life. 

The relationship was more a monologue between her and M, where she talked, and M observed. With M only replying when a direct question was asked. But when it was removed from her life, she was left with a longing she could not understand. 

Twenty years later, Dr. Alba lands at the Hölfang Foundation, hoping to find her friend and be part of the groundbreaking work at the foundation. Set in the picturesque Portugal, Alba is confronted with many questions. 

She experiences life intensely. Each person who crossed her path added another dimension to her emotional framework. This became clear when she landed in Rio de Janeiro with the promise of a beautiful romantic time with Dr. P. Which ended with her in hospital and him gone. This experience shaped her even further. 

M, the AI prototype who shadowed Alba as a child to learn about human emotions, role is very significant. Adding no spoilers to this review, the connection between the two was more than just friendship. As secrets are unearthed and files disappear, it became apparent that Alba and M’s relationship holds the key to the future of the Hölfang Foundation. 

I found the author’s writing robotic at times, and the monologues didn’t help to capture the heart of the story. Too much inner dialogue dragged the story out and hindered the flow, in my opinion. Otherwise, a good story any science reader will enjoy. 

Purchase link: 

Amazon / Universal book link / Goodreads 

Brief bio: 

Paula Hidalgo-Sanchis is a citizen of the world who has lived on four continents. She spent most of her career working with the United Nations as a humanitarian, and as a sustainable development practitioner. She led ground-breaking work on the ethical complexities, and potentials, of artificial intelligence. That work informs her debut novel Teaching Machines How to Cry. She holds a PhD in Human Geography and lives in Portugal with her two sons. 

Blurb: 

In Teaching Machines How to Cry, Paula Hidalgo-Sanchis explores the story of Alba, a young woman seeking answers for a feeling of longing she can’t explain, and M, an AI prototype. Alba was a child when she first spent time with M, until they were unexpectedly driven apart. As an adult, Alba slowly realises that M might hold the answers to her life-long yearning when she remembers that the last time she felt whole was as a child when she taught M how to cry. This novel explores the symbiosis between AI and the human soul and raises the questions of how machines and humans might live and evolve together in a not so-distant future. 

Excerpt:

Diving into humanity’s written history, the team realized that some written records, such as Egyptian hieroglyphs, weren’t easy to understand.

To address this, they adjusted the scope of the case study with a new timeframe. The new timeframe was from 2,600 B.C. onwards. 2,600 B.C. was the date of a Sumerian cuneiform script that had been identified by historians as humanity’s first intelligible written text.

Once the new timeframe for the study was set, the team combined and categorized humanity’s written legacy dating back to 2,600 B.C. Hours and hours of work were spent on collecting all available data sources, including manuscripts, historical documents, court transcripts, magazines, poetry, philosophical essays, children’s books, religious papers, song lyrics, novels, and newspapers.

Then, the scientists wrote several algorithms to do a cross-sectoral examination of livelihoods, political and societal structures, creative arts, religions, and educational systems in the texts. With computations, the team analyzed the data with scatter plots, the Pearson correlation coefficient, the correlation matrix, principal component analysis (PCA), and Lasso regression. But the computations didn’t yield the expected results, and only weak correlations and causations were found among the data points.

The findings of the case study were irrefutable: a core set of ethical values that guided people to do right or wrong in the selected timeframe wasn’t found. The team concluded that the reason was that ‘human history is too inconsistent’.

To advance the implementation of the hL-M-p, a quality assurance team proposed that E would just have to learn the seventh commandment, ‘thou shalt not kill’. The new proposal was presented to the Foundation’s board for approval. A vote was conducted. The tally was eleven votes in favor, and one against. Can you guess who that one vote came from? It was the unintentional criminal, Mr. Austin, who deeply regretted attempting to murder his mother-in-law but believed that manslaughter could be justified sometimes.

At the same time, as was the normal practice, the scientists who conducted the study on moral values published a paper with the proceedings. The paper was entitled: ‘Teaching E to be good: a scientific approach’, and it was published in the journal Science Today. When the paper was disseminated, something unforeseen happened: many voices questioned the scientific rigor of the study.

Social scientists voiced their concerns about the methods used. Historians challenged the timeframe of the case study. And anthropologists claimed that the scope of the study had missed the fact that Homo Sapiens, or the wise man, had walked the Earth for three-hundred-thousand-years. In response to the criticism, the lead data scientist issued the following statement: ‘It’s not true that the history of humanity was simplified for the purpose of the study. The broad scientific community recognizes written knowledge as a turning point in the evolution of human intelligence. The combination of logograms and syllabaries in the selected Sumerian cuneiform script is considered by many to be humanity’s first writing system. So, it is reasonable to use that readable text as the milestone that marks the beginning of the evolution of men. Humans were not evolved before they learnt how to write.’

His statement ‘humans were not evolved before they learnt how to write’ went viral and ignited more criticism. A renowned philosopher then said: ‘The Hölfang Foundation has simplified the complexity of humankind to fit it into machine-learning reality.’

After this, a media scandal developed. For the first time since it was established, the Foundation was under public scrutiny. ‘It’s an unprecedented scandal’, the chairman of the board said. The ‘unprecedented scandal’ resulted in delays in the implementation of the hL-M-p.

In addition, the costs increased by 27 percent due to additional expenditures on lawyers and marketing consultant fees. To avoid further criticism, setbacks, and extra-budgetary expenditure, the board decreed that all research would be kept secret from then on.

‘The world will still benefit from our work, but we’ll not be interrupted again’ the chairman stated. To implement this resolution, the Foundation’s board approved a confidentiality policy that all employees should follow.

After further discussions, the board decided to suspend the hL-M-p indefinitely.