Houston, We Have An AI Problem (Part 2)

control-collage b.png

Who controls the past, controls the future. Who controls the present, controls the past.
~ George Orwell

Intro 🎬

In this second installment of the series—the first one being Houston, We Have An AI Problem (Part I)—we get to explore even more strands of the cool theme of the exciting new book that is Stuart Russell’s Human Compatible: Artificial Intelligence and the Problem of Control.

Oh yes, and make no mistake: You also get to witness a whole lot more reappearances of my very own Edinburgh hoodie. Oh, the things I do for you, dear Reader.

Um, as for the underlying symbolism—the collage-format I’ve chosen to do justice in conveying this new book’s gist—nothing changes since the first installment. So in each element of the collage that immediately follows, you get to see a thematic aspect (of Human Compatible) compared and contrasted with an aspect that has stayed with me from reading a different book.

Enough of symbolism. Let’s dig—more like jack-hammer our way—right into the good (concrete) stuff.


Tis strange—but true; for truth is always strange;
Stranger than fiction.
~ Lord Byron

1. Is The Rebellion Ahoy The Real McCoy?🎈

When I first read The Robot’s Rebellion: Finding Meaning in the Age of Darwin by Keith E. Stanovich (The MIT Press) many years ago, I was struck by the neat narrative of what it means to puts human interests first. Stanovich holds the Canada Research Chair in Applied Cognitive Science at the University of Toronto, and for good reason. He writes with great originality, and he drills home the point (in The Robot’s Rebellion) that cognition is optimized at the level of genes, not of individuals. Now how about that?

Fast-forward to the newly-published Human Compatible and you get to witness a marvelous elaboration—in its own way adding uniqueness and originality—of why there’s no good reason for robots to have humanoid form. I loved how Russell points out in Human Compatible that “An accurately humanoid robot makes as much sense as a Ferrari with a top speed of five miles per hour…“.

To be sure, Human Compatible is neither an exploration of the philosophical and scientific ramifications of (Darwinian) evolution—is there any other kind?—nor a beatific meditation. Instead, it is replete with hard-hitting, carefully deliberated, and logically presented series of arguments that carefully lay out, among other things, the absurdity of elevating the status of robots.

Anyone remember Isaac Asimov’s Three Laws of Robotics? But I digress.

Anyhow, and as you’ll get to read for yourself ( in the pages of Human Compatible),  we humans had better remain vigilant to the danger that the increased use of machines in decisions affecting humans will degrade the dignity of humans. Check the delightful illustration (of this very possibility) from Russell’s reference to a scene from the science fiction movie Elysium where, in one scene, Max (Matt Damon) pleads his case before his humanoid parole officer.

And the humanoid’s response? Oh my!

We aren’t in Kansas anymore; that’s for sure.


You can’t wait for inspiration. You have to go after it with a club.
~ Jack London

2. What Inspires You? 🎪

Continuing with the theme of comparing and contrasting an aspect of Human Compatible with my impressions from reading a different book, shall we take a dip into the turbulent waters of inspiration?

When it comes to the realm of inspiration, I would say that a good place to start—assuming that you dig AI—would be the elegant monograph entitled Bio-Inspired Artificial Intelligence: Theories, Methods, and Technologies by Dario Floreano and Claudio Mattiussi, published by the venerable institution that is The MIT Press. That fine tome is squarely focused on grappling with cutting-edge approaches to AI, rooted in the worldview that intelligence is an emergent process; cells, bodies, and societies themselves self-organize in intelligence fashion, as do learning and evolution.

Let’s now segue right into the book that’s driving this essay—Human Compatible—where you’ll find its pages replete with a ton of sterling explorations into how the impulse for fostering is verily encoded in humans (think to the “caring professions”), and how that all come into play in the realm leading to inspiration. Fostering, after, is a precursor for inspiration, or so I think.


Reason: To weigh probabilities in the scales of desire.
~ Ambrose Bierce (on the word “Reason”, as defined in The Devil’s Dictionary)

3. Everything-as-a-Service (EaaS)! 🐍

Next up—and this being the third of three books that we’re hitting in turn as we compare and contrast an element or two in there with Human Compatible—my mind turned to a remarkable book I read several years ago, entitled How We Reason by Philip Johnson-Laird, published by the Oxford University Press (Oh, and don’t miss the delightful Auden quotes sprinkled throughout.)

If ever you have asked yourself, Is there a single book that has (somehow) managed to stuff between its two covers everything that you ever wanted to know about how we humans reason?, then this is your book! And I kid you not: How We Reason is equally adept at satisfying the layperson and scholar alike. In a nutshell, Johnson-Laird has somehow managed to neatly summarize pretty much all of our current knowledge and research related to demystifying human thinking.

Oh yes, you knew this was coming, just as surely as night follows day: We segue to the book that’s driving this essay—Human Compatible—and get to find out how it relates to the generality of, well, general-purpose intelligence that makes is going to make AGI possible.

Well, guess what? In case you haven’t noticed, self-driving cars have required massive investments in research and software know-how, not to mention the efforts going into addressing the enormous scope of testing all that software wizardry—algorithms and stuff like that—to make sure that they work as intended.

Yep, that’s good old engineering at work.

Speaking of which—software engineering to be precise—you are definitely going to recognize some of the buzzwords that abound here. To name but a few: SaaS, IaaS, and PaaS. Well, guess what, check out Russell’s fascinating take on evermore intriguing acronyms. To name but a few: TaaS, EaaS, etc. (I particularly liked the last one, EaaS. Oh yeah, that would be Everything as a Service.)

Get ready to be regaled with a mind-expanding narrative of how, starting with us humans getting used to the idea of Transportation as a Service (TaaS), and how we are inexorably headed for—yep, you guessed it—general-purpose AI (aka Everything as a service (EaaS).

At the embodiment level, general-purpose AI would have free rein to the entire knowledge and skills—in all their varieties—of the human race, and then some. That, dare I say, makes reading Russell’s Human Compatible a splendid follow up to Johnson-Laird’s neat summarization of pretty much all of our current knowledge and research, as they (I should add) relate to demystifying human thinking.

He who confronts the paradoxical exposes himself to reality.
~ Friedrich Dürrenmatt

The End 📺

And so it is that we wind down our journey. Wait, what?!

That gondola again!! Folks, while we figure out this freak occurrence, rest assured that we’ve got one more installment coming up for you: Saving the best for the (forthcoming) third of the three essays which commemorate Human Compatible.

boat b.png

Your Comment Here!

This site uses Akismet to reduce spam. Learn how your comment data is processed.