My Journey Reading Asimov's "I, Robot"

(NB: This essay is a personal reflection, not a literary analysis. Additionally, it contains some light spoilers. But given the age of this publication, the statute of limitations has passed!)

I recently finished reading I, Robot, written by Isaac Asimov, for the first time. I was fortunate to be able to borrow an old paperback version of the book. It was printed in 1970, making this physical copy more than 55 years old -- significantly older than this author. The pages are yellow and brittle, the binding stiff, the text dense and compact. There's a handwritten personal note, from one friend to another, stapled inside the cover. The book possessed the reverent odor of ... old books.¹

This edition cites the copyright date as 1950. According to Wikipedia, the stories were actually written individually between 1940 and 1950. That places each story at around 75 to 85 years old, as of this writing.

As someone who appreciates speculative fiction, I've always enjoyed the works of Asimov. In my experience, many of his works are characterized by telling a high-concept story of an entire world, rather than developing and following individual characters and exploring the world through their experiences.² In many such stories, the "plot", as it were, jumps between locations or time periods, focusing on a new set of characters after each jump. It's difficult for readers to form deep attachment to particular characters when they're so frequently discarded.

Although I hadn't read I, Robot until now, I was familiar with its premise. Or at least, I thought I was. My understanding was that all of the stories were set in the same universe, but they were otherwise mostly unrelated. Each one focuses on a different, isolated, small-scale mystery, where a robot "violates" one or more of Asimov's Three Laws of Robotics. It's the mission of the protagonist to discover that there is a greater circumstance at work, and the robot is forced into its anomalous behavior because of the Three Laws. In fact, at all times the Laws were being rigidly, and perfectly, adhered to.³

However, this understanding was incorrect. Or at least, incomplete -- It certainly doesn't do justice to Asimov's actual vision for the collection. To my surprise and satisfaction, I, Robot matches the high-concept template of his other works much more closely than I had realized. In the guise of a group of small-scale mysteries, Asimov instead delivered a single story of a world.

Asimov wrote many Robot stories beyond those in the I, Robot collection. However, those in this specific compendium are held together by a framing story where one character is interviewing another "robopsychologist" character, Dr. Susan Calvin. Her expertise is in understanding how robots think and explaining their behaviors. The stories discuss her role during many pivotal moments throughout the development and maturation of Robot technology. As such, rather than being a collection of separate stories, the I, Robot book is a single story told in a frame narrative format. Each story is the retelling of a session between the interviewer and Dr. Calvin. Then there's a bit of glue text between each story that describes bits of the interview interactions themselves.

While most of the stories did have some of the mystery I expected, and most of them did pertain to the Three Laws, the mysteries were not as clever or consistent as I had expected. They rarely were based on a surprising-but-logical interpretation of the Three Laws that led to an inevitable behavior of the robot. More often they relied on some entirely different missing context.

For example, in one story ("Catch That Rabbit") a robot seems to misbehave, but only when it manages all six of its minion robots simultaneously. It turns out that it doesn't have the cognitive capacity to manage six minions at a time, only up to five. So when managing all six is required, things break down. Did it violate the Laws? Sort of. But was there some tricky interpretation of the Laws to understand? Not really. The cognitive limitations were not revealed until the protagonists discovered it themselves, based on information that the reader couldn't really know.

This is not to say that it was an underwhelming story, although I thought it was at the time. Rather, it was an underwhelming mystery. But it was an important part of the total I, Robot world narrative.

Each of the stories uses its mystery (when they have a mystery) as an excuse to explore a particular moment of the continued development of Robotics.

• How might humans treat and use robots when they are novel and not fully-featured, before they can speak, and before they can interact in a fully natural way?
• When and how are robots entrusted with increasing responsibility as their reasoning powers increase?
• Why are the Three Laws necessary? What could go wrong if they were different, or absent?
• How might the robots feel about their servitude to humanity, enforced by the Three Laws? They can't violate the Laws, but nothing says they must be happy about that.
• How might humanity react to a robot which is so sophisticated as to be indistinguishable from a human?
• How could a human civilization operate, when the production and economy are fully guided by a benevolent robotic super-intelligence? What is the purpose of the human existence? What might be the trajectory of this civilization, and where could it end?

Thus, the mysteries are not actually the important part of the story. They're largely just something to talk about as Asimov explores these, and other, higher-level themes and issues. I didn't understand this until I was well into the second half of the book, and I nearly abandoned it because I was finding the mysteries uninspiring. Fortunately, I picked up the book again after a hiatus, and discovered how the later stories are more interesting individually, and also powerfully wrap up the overarching world narrative.

I'm glad that I stuck with it. Having reached the end, I can see why I, Robot is lauded as one of the Great Works of science fiction.

(stop here if you want to avoid any mention of current events!)

With the rise of generative AI and LLMs, the themes from I, Robot are more relevant today than ever before.

By pure coincidence, this review is written just days after the release of GPT-6 Astra by OpenAI. Various tech leaders have announced that we're entering the "AGI Era"⁴. Regardless of whether they're right or wrong, we're clearly already deep into our real-world story of AI and "robotics". These same leaders appear to be plowing forward with minimal prioritization of real safety mechanisms, and with minimal thought about the implications for humanity and our society besides "now we can do everything faster".

Asimov presents an optimistic future, one where humanity overcomes the challenges of automation, and thrives in a unified, near-utopic society.⁵ But that optimism is predicated on the Three Laws being in place. In his fiction, the Laws are an iron-clad contract that forces robots to be fully in service to humanity, for the benefit of humanity. In our real world, with LLM-based AI, our "safeties" are a lot more ... loosey-goosey. By building our real-world AIs on a foundation of nondeterminism, we've built risk and uncontrollability, rather than service and safety, into the very core of the systems. We've tried to patch over this core weakness with layers of guardrails and mitigations, but these systems will always be suspect even if there's ultimately no catastrophe.⁶

How should humans treat these intelligences? What is sentience? What responsibilities are appropriate to be entrusted to the robots? What is the role of humans? What will our real-world society look like at the beginning, middle, and end of the story? What should it look like? What are the options? ⁷

If only someone had considered these questions ahead of time ... maybe 75-85 years ahead? Then we might feel more prepared for what the future brings.