Books I Read Each Year
Transcendent Kingdom by Yaa Gyasi (January) My Robot Gets Me by Carla Diana (January)
The Meaning of Human Existence by Edward O. Wilson (December) The Midnight Library by Matt Haig (December) Never Let Me Go by Kazuo Ishiguro (December) Snow Country by Yasunari Kawabata (September) what to expect when you're expecting robots by Laura Major and Julie Shah (September) Klara and the Sun by Kazuo Ishiguro (September) Recursion by Blake Crouch (September) Incognito: The Secret Lives of The Brain by David Eagleman (September) Circe by Madeline Miller (June) The One World Schoolhouse by Sal Khan (June) Deep Work by Cal Newport (June) Consciousness and the Social Brain by Michael Graziano (May)
Dark Matter by Blake Crouch (November) Dune by Frank Herbert (June) Baptism of Fire (Witcher Book 3) by Andrzej Sapkowski (May) The Time of Contempt (Witcher Book 2) by Andrzej Sapkowski (January) Superintelligence by Nick Bostrom (January)
By Yaa Gyasi
Date Finished: Jan 9, 2022
This is one of the best books I’ve read in a long time. It deeply discusses the meaning of life, addiction, God, the importance of kith and kin, poverty, racism, and much more. It’s about Gifty, a woman who is a first generation American of Ghanian descent. She’s now doing her PhD at Stanford in neuroscience, studying the reward seeking behavior of mice. The author does an amazing job utilizing memories of Gifty’s past by jumping around different parts of her life to explore her thoughts on various topics. I found a lot of it particularly relatable since Gifty grew up in a low-income, single mother immigrant household. Because Gifty came from vastly different experiences from those she was around, it became somewhat isolating for her. I’ve definitely felt this way before, and this book does a brilliant job expressing these emotions.
Although I’m not very religious, I found it interesting how Gifty reconciles her belief in God over time in contrast to her views as a scientist, which the author excellently portrays by going back and forth from childhood to adulthood. Shifting from past-to-present really helps hit home how her ideas changed on science and religion in childhood, undergrad, and the present-day.
By Carla Diana
Date Finished: Jan 9, 2022
This book focuses on product design, and uses ideas in interaction research to help design better products. The author of this book is the Head of Design at Andrea Thomaz’s Diligent Robotics. Since robots most HRI researchers work on tend to be relatively social, Carla Diana decomposes it to understand how to create well-designed interactive products. She proposes a framework composed of Presence (the form of a product/robot), Expression (how it communicates using sound, light, motion, etc), Interaction (using sensor data for an interaction feedback loop), Context (social contexts like time/place to influence interaction), and Ecosystems (how behaviors can work together across skills). Although I agree with the design of the framework, some of the descriptions of the products as examples of good “social” design seemed like a stretch to me.
I found the overview pretty nice, but I the description of some of the product-specific examples seemed to vary in depth. Some were excellently discussed, while others had a few paragraphs of discussion that did not add much to the overall arguments. I do think the book does do a good job talking to its target audience of people who actively work on product design on how human-centered social design can play a more impactful role.
By Edward O. Wilson
Date Finished: Dec 28, 2021
I originally picked up this book for UT’s Annual Reading Round-Up for freshmen, but never finished it after my discussion group with a professor got cancelled. As I started rereading the book, the author, famed biologist Edward O. Wilson who is often called “Darwin’s heir”, passed away. The book is generally pretty interesting, and since Wilson’s most famous work is on ants, there’s a lot to learn about social evolution from them (thought most of it not practical for humans to apply to their personal lives). It’s a pretty short read which recaps how humanity potentially evolved to our dominance on Earth, and how we should not enter the Anthropocene, an age of history where only humans matter. We should protect biodiversity on our planet rather than focusing only on our own existence.
Wilson ponders what the meaning of human existence is. He explores that the meaning comes from the “accidents of history, not the intentions of a designer”. He states that the humanities describes the human condition, while science encompoasses the more general meaning of human existence. He argues that the meaning of human existence then comes down to what created the human species, and that it was random events during evolution led to where we are today, not predestination. Wilson evaluates the nuances to natural selection and evolution to examine the “Anthropocene Epoch”, detailing how humanity’s social behaviors are what differentiated us from most other species.
In particular, he contrasts kin selection, where an individual being altruistic tends to benefit its close relatives (and thus preserves some of their genes) to multilevel selection, where natural selection occurs between individuals within a group and between groups. Wilson is one of the originators of multilevel selection, and claims that this belief has been gaining popularity, though does have staunch detractors such as Richard Dawkins. Wilson claims that this multilevel selection enhanced the social intelligence of Homo sapiens, and led to our dominance on earth. Unrelated, but this work was what motivated genetic and evolutionary algorithms which I previously worked on, so it was interesting to read about the biological origins and debates that took place to develop these theories.
Whereas other animals have amazing sensory capabities, we need instruments to measure most forms of stimuli. Humans, however, are narcistic about themselves, and this anthropocentricity lets almost every human be a genius in understanding complex social cues to cooperate, bond, gossip, and control. We even create animal caricatures to attach meaning human-like emotions to non-human entities.
By Matt Haig
Date Finished: Dec 27, 2021
I started reading this book because I saw it won Goodreads Choice Award for Fiction in 2020. The book follows Nora Seed who after having so many bad experiences and feeling unneeded, decides to give up on life. She finds herself in The Midnight Library, where she can experience the infinite possibilities of how her life could have been.
This book takes the “what-if” style of thinking to the extreme, similar to Dark Matter by Blake Crouch. This book is just Dark Matter but trying to find personal happiness rather than being more plot-heavy. I do think I enjoyed Dark Matter more than this book, but they are ultimately different things.
The book is okay, but I wouldn’t describe it as good or great. It’s a pretty fast read because of the simple style of writing, so I finished it in 2 days. It’s an enjoyable way to spend some free time, but I’m not sure how it won the Goodreads Choice Award for Fiction to be honest.
The main character studied philosophy in her main life, so throughout the book, tidbits of philosophy are just dumped. Usually it’s a bit on the nose with a lot of telling and not showing. It’s a little much if you start quoting the Robert Frost’s overused, and often misunderstood, The Road Not Taken poem. With all of the quotes to Thoreau and Plato, I was hoping for a reference to Jorge Luis Borges’ The Garden of Forking Paths, which preceded the many worlds interpretation the novel relies on.
By Kazuo Ishiguro
Date Finished: Dec 25, 2021
This book really highlights Ishiguro’s stream-of-conscious writing style and use of unreliable narration that was also in Klara and the Sun. This book follows the experiences of Kathy, Ruth, and Tommy who grew up in this boarding school called Hailsham. We are told early on that the students at this school are meant to be carers, but for who and why largely unknown for most of the book. This style of narration slowly exposes us to the slightly dystpoian nature of the world.
Without spoiling anything, I would describe Klara and the Sun to be the spiritual successor to this book, since it focuses largely on the same themes such as the pursuit of art and the importance of childhood. I would personally recommend this book, but I can see others not liking Ishiguro’s writing since largely nothing happens plot-wise. Ishiguro focuses mostly on the development of the characters, and uses realizations in world building to test the characters.
By Yasunari Kawabata
Date Finished: Sep 20, 2021
Yasunari Kawabata won the Nobel Prize in Literature in 1968, and was the first Japanese winner of the award. I read his short story “The Dancing Girl of Izu” and enjoyed it a lot, so I picked up this book. At a high level, this book is about wasted love. The main character, a married man named Shimamura, visits a hot-springs town and meets a local geisha, Komako. It is established early on that city geisha have real influence, but hot-springs geisha struggle to survive and often border being prostitutes. Thus, it’s common for these geisha to find a man to fall in love with them, so they can get out.
Komako genuinely falls in love with Shimamura; however, Shimamura distances himself from his emotions. Kawabata’s fame comes from his amazing ability to use implicit imagery to portray the characters. Komako is often depicted as the color that contrats against the overwhelmingly large, blank, white snow. Since much of the story is not expressed directly, the book is a bit difficult to read, but it is still a great short novel to read in a sitting or two.
By Laura Major and Julie Shah
Date Finished: Sep 16, 2021
This book, by Laura Major and Julie Shah shows the reality in designing robot systems that are regularly around humans. I started reading this book before I started my PhD, so it gave me a lot of interesting ideas to look into as I started ramping up at USC.
The book interlaces many of its points with stories in aviation and human-robot interaction research, making for an enjoyable reading experience. Since I’m still finding my bearings in HRI research, this book provided a really good introduction to the effects of the design decisions that researchers study. The emphasis that people come from different cultures and its relationship to the social acceptance of autonomous robots really hits it home for me. Additionally, since robots have their own goals, and there are likely going to be many competing companies with their own robots, the interaction complexity with humans skyrockets. I’ll probably be coming back to this book in the future.
By Kazuo Ishiguro
Date Finished: Sep 16, 2021
Kazuo Ishiguro’s writing in the perspective of Klara, an Artificial Friend (AF) who has little knowledge of the outside world, is a great way to slowly introduce the reader to how the world works. Very rarely does Ishiguro explicitly tell the reader what’s going on. Instead, he gently implies many elments of the world from the skewed perspective of Klara. I think this book investigates many different kinds of love between characters, and Klara’s child-like understanding of the world gives us a fresh perspective on these themes.
I actually picked up this book not knowing it was about AI. Since this book is about a robot that can understand social interactions, it turned out to be perfect for me. Several times throughout the book, it mentions how to gaze at other people in group settings and how to portray oneself, which is important from an interaction perspective. Ishiguro apparently met with the Demis Hassabis (co-founder of DeepMind) to ensure that the AI-perspectives were accurate within the novel. The book briefly comments on the general social acceptance of realistic, social interactants like Klara, which is something not many HRI researchers consider from my understanding. As we learn more about the world, it feels somewhat dystopian, but I do think Ishiguro is an optimist about AI, gene-editing, and other futuristic inventions. I might give this book another read one day to pick out the tiny details I missed.
By Blake Crouch
Date Finished: Sep 14, 2021
After reading Blake Crouch’s Dark Matter, I decided to follow it up with this book. The style of plot is pretty similar between this book and Dark Matter. I still enjoyed the sci-fi concepts in this book, since it felt like a refreshing new take on the genre (at least in my limited experience). I don’t want to say too much since it might spoil it, but this book was a good read, similar to Blake Crouch’s Dark Matter. I think the character development was weaker in comparison to Dark Matter, but it was explained in a way where it was reasonable to skip the “development”. Similar to Dark Matter, Crouch’s distinct writing style, along with the intensive plot, makes it incredibly easy to finish the entire book in one sitting.
By David Eagleman
Date Finished: Sep 7, 2021
In this book, David Eagleman provides a broad overview of different functions in the brain. The book is pretty interesting, with many high level stories of the oddities which exists in neuroscience. I found certain sections more interesting than others. I would definitely recommend it to anyone interested in a high-level overview about cool things in the brain.
Eagleman claims that the brain is a team of rivals. Within these rivals can exist, for example, more racist versions of yourself, along with more rational sections (explained using an example of Mel Gibson’s racist remarks during a DUI). He claims that sometimes your brain can lean one way or the other, but overall, the majority solution of the brain leads to your decisions. He provides evidence through split-brain examples, where some people have one hand buttoning up a shirt, while the other unbuttons or fights the conscious part of the brain.
This resembles Marvin Minsky’s thought’s on the topic: that the human mind consists of a large number of small, interacting subagents. These agents build up to create a “society of mind”, and competition within the society creates your decisions. Because of these examples, Eagleman believs a team-of-rivals framework trained in an evolutionary appoach will lead to a new age of bio-inspired AI. My qualms with this is that training through evolutionary means is extremeley compute inefficient, especially if you’re attempting to optimize both a team-of-rivals and various types of subagents.
Eagleman also questions the legal system’s assumption that all people are equal. He argues the law is pretending that all brains are equal, when they are definitly not. Currently, if you’re 18 or have an IQ of 70, you get the death penalty, but if you’re one day away from 18 or have an IQ of 69, you live. These are crude ways to distinguish when someone isn’t fully in control of their actions; however, many undiagnosed mental or genetic correlations have been found in certain crime categories, many of which are potentially treatable. Eagleman thus proposes a personalization of the law such that the punishment is aligned with neuroscience. He states we should throw away the biased intuitions about blameworthiness with a fairer approach. Although these sound agreeable, he also pushes for warehousing people who will not be changed by punishment. This feels low-key dystopian, and reminds me of the anime Psycho Pass (which I watched when I was a kid), where those who were more inclined to commit crimes were imprisoned before any crime had been commited. My primary issue with these kinds of approaches is that it’s probably difficult to guarantee without a doubt someone is unlikely to change, thus dishing out varying level of punishments are somewhat unfair.
At the end of the book, Eagleman makes an argument against reductionism. He states that when systems begin to have many smaller parts, you should stop looking into the tiniest of granularities, but rather focus on the larger pieces. The interaction of these larger pieces can create emergent properties which can create more relevant outputs. I feel like in the realm of AI resarch, there has been a slight shift away from reductionism (building low-level models). Instead we’re focusing on the presence of large amounts of data (which Yuval Noah Harari labels as dataism in Homo Deus), and modifying existing models to work with the data.
By Madeline Miller
Date Finished: Jun 28, 2021
This book is about Circe, the daughter of the Titan Helios. This book reimagines the Circe from The Odyssey, and portrays her as relatable. Circe lived a lonely life in the beginning of the book, where she encounters several famous mythological figures such as Prometheus, Daedalus, and others. I found the earlier part of the book slow, but it helped define her character as she got into witchcraft. After she was banished to the island of Aiaia, the story quickly speeds up. I really liked how Circe had to put lots of hard work to master her craft rather than being born great like her siblings.
How the author wrote Circe’s inner thoughts was great, especially when she was in dialogue with other characters. She always reflected on the relationships between gods, her thoughts on how people behave, and more. It helped ground Circe to humanity rather than divinity, even when she had to present herself more formally.
By Sal Khan
Date Finished: Jun 24, 2021
Sal Khan, the founder of Khan Academy is concerned with the modern education system, which originated from the Prussian system of education. Many students struggle in the education system, as they don’t learn in any depth what they are learning, mostly due to inconsistent testing and scale. Sal proposes a “One World Schoolhouse” as a combination of high-quality at-your-own-pace online tutorials followed by assistance from many teachers in a mixed-age classroom.
The current semi-standard education system in the US came to be to largely educate the populace, but its origins from the Prussian system are somewhat interesting:
His motivations were generally forward-thinking for the time; he wanted to provide a solid basic education to students of all socioeconomic ranks. As in Prussia, this would play a significant role in building a middle class capable of filling the jobs of a booming industrial sector. There was, however, also an element of indoctrination that had positives and negatives depending on your point of view. While it would be beyond the scope of this book to examine in detail the political climate of the time, suffice it to say that in the 1840s—as today—the United States was faced with the issue of “Americanizing” large groups of immigrants from many disparate cultures.
He generally brings up flaws in the current education system that I do think need to be resolved. The current education system doesn’t help everyone, just some people. I do think the most stellar students will succeed anyways, but our current system separates out the “gifted” students from the “average” students. He says:
This means putting the “fastest” students in “advanced” or “gifted” classes, the average students in “average” classes, and the slowest students into “remedial” classes. It seems logical… except for the fact that it creates a somewhat permanent intellectual and social division between students.
His solution is to use Khan Academy (or some future tool) to make education more accessible and at the pace of the student. In addition, he wants to throw away the rigid structure of schools, and instead have a mixed-age classrooms with many subjects being taught at once. To do this, there would be many teachers from all sorts of backgrounds to help students succeed in their interests, as well as students helping each other. This approach likely has many benefits, but I worry about effective testing strategies, or approaches to measuring this success. Also, how would you handle students not context-switching effectively? How do you handle the students who are transitioning between the old rigid structure to the new structure? There are many questions without answers, but I find this approach to education promising.
He also states the following rebuttal to a criticism that I had:
I sometimes get pushback from people saying, “Well, this is all well and good, but it will only work for motivated students.” And they say it assuming that maybe 20 percent of students fall into that category. I probably would have agreed with them seven years ago, based on what I’d seen in my own experience with the traditional academic model. When I first started making videos, I thought I was making them only for some subset of students who cared—like my cousins or younger versions of myself. What was truly startling was the reception the lessons received from students whom people had given up on, and who were about to give up on themselves.
I disagree with this, and I still do think that nonmotivated students would struggle or be distracted in environments such as these. His experiments tended to happen in richer schools where funding was high. Students with, say a bad home environment, can be adversarial or difficult, and are often relegated to the worst of schools (in regards to funding and education quality). How do we ensure that these students can also succeed when everything has already been failing them? I do think it’s more of an infrastructural issue that needs to be jointly solved.
One of my favorite quotes in the book:
At the end of the day, however, the fact is that we educate ourselves. We learn, first of all, by deciding to learn, by committing to learning. This commitment allows, in turn, for concentration. Concentration pertains not only to the immediate task at hand but to all the many associations that surround it. All of these processes are active and deeply personal; all involve the acceptance of responsibility. Education doesn’t happen out in the ether, and it doesn’t happen in the empty space between the teacher’s lips and the students’ ears; it happens in the individual brains of each of us.
By Cal Newport
Date Finished: Jun 15, 2021
Since I’m an incoming PhD student in computer science, I felt this book had a lot of little gems of advice for succeeding in grad school. The general premise is simple: you should work in deep concentration in a distraction-free environment. You should avoid shallow work like logistical tasks (e.g. email) as much as possible. By working deeply for longer, you’ll be much more creative and productive. I found Cal Newport’s four disciplines to be super helpful:
- Discipline 1: Focus on the Wildly Important. You should focus on a small number of “wildly important goals”. This could be the number of papers you want to publish or the field you want to be an expert in.
- Discipline 2: Act on the Lead Measures. You need to have a way to measure success. There are lag measures and lead measures. Lag measures describe what you want to improve. When you recieve a lag measure, it’s too late to change your behavior since it’s already happened. Lead measures “measure the new behaviors that will drive success on the lag measures.” A lag measure could be “how many papers you published per year”, while a lead measure that’s more effective is “how much time you spent in deep work”.
- Discipline 3: Keep a compelling scoreboard. Track your lead measures such as logging how many hours you worked deeply each day.
- Discipline 4: Create a cadence of accountability. Review your scores and understand why you’re doing better or worse, and act on those results.
I have been trying to incorporate some of these into my own life by scheduling out chunks of time to only work deeply. I would say this is a must-read for any incoming PhD student. Since Newport is a professor, he pulls a lot from his experiences in academia, making much of it relatable. Even though the core idea of the book is obvious, the anecdotes and evidence Newport provides is motivating.
By Michael Graziano
Date Finished: May 9, 2021
This book was for one of my classes this semester, and I really enjoyed the ideas tossed around this book. It provides a great introduction for people who don’t know anything about this space The brain is one of the most complex information processing machines in existence. Existing research in neuroscience tells us that the electrical signals in the brain essentially move from region to region, and the signal strength is boosted, improved, or maintained, but the key question is how does this become the subjective experience we as humans feel every day. This is ultimately a question on what is consciousness. Graziano proposes the atention schema theory of consciousness as a simple approach for addressing this question.
Before getting to Dr. Graziano’s theory, two key parts must first be understood. First, is the concept of attention, which can intuitively be thought of as the brain putting its focus on certain signals over others. This is an information processing trick to reduce a large number of signals to its most significant pieces. Second is the idea that the brain creates simplified models of objects and events in the world, and these models are then used for making predictions, making actions, and reasoning about the world. In addition, a relatively formal definition of consciousness is needed, which Dr. Graziano explains as the “whole of personal experience at any moment.”
What makes attention schema theory interesting for me is the fact that we generate schema not only for ourselves but for other people and things (living and nonliving). These schemas are quick-and-dirty models we construct and impose onto other entities. For example, a robot playing soccer doesn’t have any actual emotions, consciousness, or awareness, yet we still impose some model onto them (potentially attributing some human-like characteristics to it). Applying an attention schema onto ourselves gives rise to self-awareness, whereas on others gives rise to social awareness.
Dr. Graziano also discusses the building of conscious artificial intelligence; he states that it requires three things: to be able to sort information and control its own behavior using attention; to be able to run its own attention schema, tracking, simulating, and predicting attention; and it needs to be able to link its attention schema to other information it has stored. For conscious AI to exist, it will need to create an attention schema of itself, but to do this, I think it would need some method of communication with the world, such as language, robot gestures, or robot gaze, and also be able to model the effects of its communication with the world.
I believe that Dr. Graziano does a wonderful job creating a simplistic, yet powerful theory of consciousness. His theory has solid foundations in plenty of empirical research, which provides for a compelling argument. His comparisons to existing theories such as social theories and information-based theories were particularly interesting and gives people a general overview of the space. I recommend this book as a good introduction for anyone interested in consciousness in general.
By Blake Crouch
Date Finished: Nov 27, 2020
I hadn’t read for almost half a year until I started reading this book. It’s almost impossible to put this book down once you start, and I finished it in two sittings. Blake Crouch writes in a simple way with single-line paragraphs and sentence fragments to focus on the fast-paced nature of the book. The story follows Jason Dessen, a small-time college physics professor who one day gets abducted. I think this is the kind of book that should be read without knowing anything about it beforehand besides what I have already said.
The stakes are always high, and I was always on the edge of my seat. I highly recommend this book for anyone who wants to get back into reading since it’s easy to get hooked.
By Frank Herbert
Date Finished: Jun 12, 2020
There’s not much else to say about this book that’s not already known. There was a resurgence in popularity of Dune since they upcoming release of the new movie. This book is long, and especially since I was trying to get back into reading again, this took a while to get through. The book is set in a future where a planet called Arrakis contains a drug known as “melange” or “spice”. In this world, “thinking machines” were thrown away long ago, and instead spice is used to enhance pilots during space travel. The plot follows the young heir of the House Atreides, Paul. His father, duke Leto, was given Arrakis by the former owners and enemy of House Atreides: House Harkonnen.
This book set the foundations for much of science fiction today. It has political scheming, imperialism, religion, and interesting takes on the role of augmenting human intelligence in the future. A lot of the terminology in the book are Arabic and come from Islam, which is interesting. I feel like the Middle East was relatively unknown to people in the 1960s, so there probably was somewhat of a novelty factor at the time. The politics within the novel is allegorical to the Middle East, where “spice” can effectively be replaced with “oil”.
Some argue that there is some religious approriation going on here, but I don’t think the book portrayed anything negatively. It is interesting to see how Islam influences his world though. Since the book was written in a pre-9/11 world, so it’ll be interesting to see how the new movie will translate some of the book to the movie, since it seems they’re potentially rebranding “jihad” to “crusade”.
By Andrzej Sapkowski
Date Finished: May 15, 2020
I really enjoyed this book and I definitely enjoyed it more than the other books in the series. It had a nice cast of enjoyable characters, some new (Regis) and some old (Dandelion). Geralt thinks Ciri is in Nilfgaard and starts to journey there, but Ciri is really with the Rats she joined at the end of the last book.
Geralt ended in Brokilon with Milva and Dandelion. They start going towards Nilfgaard. Geralt ends up bumping into Zoltan Chivay and his group of dwarves who were guiding refugees from the war (they seem like they’re nice people but at the same time Zoltan claims he’s not virtuous as he’s still a simple theif robbing people). Cahir, the Nilfgaardian officer who’d been following them, ends up joining the group. Also Regis, an odd barber-surgeon, they found along the way joins the group. This little posse creates a nice road-trip vibe as they travel together.
There was this funny moment where a vampire character talks about how he quit drinking blood since blood is like alcohol and how he’d go into these crazy benders and parties with other vampires. He broke up with his vampire girlfriend and he became a bigger bloodoholic. One day he was “flying” under the influence and crashed into a well. He was then captured and buried alive by humans. He then decided to quit drinking blood. I just found it hilarious they had a story of an “alcoholic” vampire.
Sapkowski has a way with exploring philisophical ideas of choice and morality through dialogue between Geralt and other characters. With Nilfgaard at war with basically everyone else, Geralt is stuck with many questions where he must make choices whose answers are not black and white. The dialogue the characters engage in is creatively filled with humor and friendly banter while still tackling serious questions. This think this book is the strongest contender for my favorite book in the series so far, and it definitely makes the series worth reading.
By Andrzej Sapkowski
Date Finished: Jan 9, 2020
I started reading the Witcher books in 2017 and 2018 after playing the wonderful video games that take place after the franchise. This is the fourth book in the Witcher franchise, but is the second novel that picks up from the Blood of Eleves. The first two books consist of short stories which build up the world of the Witcher. Compared the previous Witcher books, there’s a lot more political turmoil and plotting, and much less monster hunting.
It was really cool to see many new characters who are familiar in the video games being introduced in this book, but I think if I hadn’t played the games and been familiar with the world, it would have been difficult to know who rules what kingdom and why each kingdom hates one another. This book wasn’t as good as Blood of Elves (Book 1) or Baptism of Fire (Book 3), but it still builds a great world with lots of political scheming.
By Nick Bostrom
Date Finished: Jan 6, 2020
The book itself is hard to read as it’s dense with examples, but is a good read. The book has a lot of great snippets, but these are the parts I enjoyed the most.
Bostrom claims there are a few ways to achieve superintelligence, and achieving any of them will lead to any other. One can use artificial intelligence, emulate the entire brain, develop the human brain to be a weak form of super intelligence (which could then create stronger forms), and a few other methods of creation. Once a superintelligence is created, it could be sped up with excess computational power, expanded using many smaller intelligent units to collectively make decisions, or be exactly like the human brain in terms of speed, but be able to make better decisions/calculations than humans can.
He explains that having one single superintelligence that’s friendly towards humans is good towards us, but if it ever turns against humanity or makes a decision not in our favor, it could have catastrophic outcomes. Bostrom discusses numerous potential protections for such an issue.
Bostrom asks what would happen if the development of superintelligent systems was multipolar; where many superintelligent systems could exist at any point (perhaps developed by different countries or companies), thus a single agent doesn’t take over the world. Rather, a bunch of agents can exist at once. Through this Bostrom explains a potential algorithmic economy, where the laborers can be superintelligent human-like agents that can exist for days at a time in the digital world. Bostrom here is slightly optimistic I think on this view of what would happen in this case, as the rich who own significant capital can take advantage of the effects of the economy to become wealthier, but those who don’t have any capital would likely need to be helped through philanthropic means from those who became wealthy (I think this is unlikely because look at the billionaires today). Bostrom talks about Malthusian conditions, where in the past, the growth of societies were to an extent limited by the amount of resources one had, but we’ve never reached a point to use all our resources at a maximum consumption rate. Theoretically, catastrophes and plagues won’t have to help limit population in a society that’s more or less controlled by AI. The issues with a multipolar system is outcomes can be uncertain, as game-theoretic approaches might be needed to make the best decisions given all the other agents that exist at the same time.
He also discusses the strategic development of AI for the greater good. It essentially says that the development of a singleton is the best given the proper solutions to the control problem exist. It also discusses the issue of collaboration to build a singleton rather than have many different organizations build their own superintelligences which create the multipolar problem. At a certain point, Bostrom becomes very optimistic, saying that a rule such that after a certain amount of money is made, any excess can be distributed across all of humanity in order to keep the rich from getting richer from having full access to the computational powers of AI. It does bring up an interesting thing of how much of a superintelligence you own can define how much you are worth. But Bostrom’s idea of democratizing parts of a superintelligence to benefit humanity, even though I find the idea great, I feel as if it would never happen due to the greediness of those who own the superintelligence.
Bostrom gives a nice analogy for AI today: humans messing with AI is the same as giving a child a bomb to play with. Any sensible person would simply put down the bomb, but being children that’s not always the case. Now, instead of one child, there are many, and if any one of them blows up the bomb, it messes everything up.
AI ethics is a real issue, and many companies are slowly trying to look into it (except when companies like Google terminating members of their ethical AI teams). One issue that Bostrom talks about is which we solve first, the control problem or superintelligence. The fact is, we are going to advance towards the latter as fast as possible, thus the control problem may not be solved first, causing potentially disastrous results. For this reason we must look into this problem now.