AI ebook cover design brainstorming
Thanks to everyone who brainstormed possible titles for MIRI’s upcoming ebook on machine intelligence. Our leading contender for the book title is Smarter than Us: The Rise of Machine Intelligence.
What we need now are suggestions for a book cover design. AI is hard to depict without falling back on cliches, such as a brain image mixed with computer circuitry, a humanoid robot, HAL, an imitation of Creation of Adam with human and robot fingers touching, or an imitation of March of Progress with an AI at the far right.
A few ideas/examples:
-
Something that conveys ‘AI’ in the middle (a computer screen? a server tower?) connected by arrow/wires/something to various ‘skills/actions/influences’, like giving a speech, flying unmanned spacecraft, doing science, predicting the stock market, etc., in an attempt to convey the diverse superpowers of a machine intelligence.
-
A more minimalist text-only cover.
-
A fairly minimal cover with just an ominous-looking server rack in the middle, with a few blinking lights and submerged in darkness around it. A bit like this cover.
-
Similar to the above, except a server farm along the bottom fading into the background, with a frame composition similar to this.
-
A darkened, machine-gunned room with a laptop sitting alone on a desk, displaying the text of the title on the screen. (This is the scene from the first chapter, about a Terminator who encounters an unthreatening-looking laptop which ends up being way more powerful and dangerous than the Terminator because it is more intelligent.)
Alex Vermeer sketched the first four of these ideas:
Some general inspiration may be found here.
We think we want something kinda dramatic, rather than cartoony, but less epic and unbelievable than the Facing the Intelligence Explosion cover.
Thoughts?
Planet Earth, turning into paperclips. This has the potential to become an iconic image.
If this were a different type of website, it would have long ago. And someone role-playing an AI would use an 80x80 version as their avatar.
(This suggestion is the only one in this discussion so fa that’s gotten my attention. However, I’ve already been exposed to the paperclip death of the universe scenario, so not sure how it would work on the unfamiliar.)
I don’t think it would work all that well on the unfamiliar.
Yes, for newcomers it would be unfamiliar. It’s not so crass as to look like an attention-grabbing ploy—just strange enough to pique curiosity.
Possible alternate version: An entire city given over to paperclip manufacturing.
A human and something much less intelligent than a human.
I like this. For example, a human holding a white mouse in one hand and looking at it.
The problem, though, is that once you have a human in the shot it’s real hard to get someone looking at the book for the first time to identify with the white mouse. One strategy is to de-humanize the human—like maybe only have a human eye filling up the background, with the shot centered on a butterfly. example, sort of. Eh, seems pretty difficult now that I think about it more.
A mouse, a human, and a server rack, each bigger than the next, side by side against a black background. Scaled so that the difference in size between the mouse and the human is the same as the difference between the human and the server.
I like the idea of suggesting the recursion of human intelligence to machine superintelligence.
How about a hunter-gatherer working on making a spear or other primitive weapon, or maybe fire, in the foreground, and an ambiguously human-robot-cyborg figure in the background with a machine gun / nuclear generator?
And if you look carefully at the floor the human is standing on… it’s a robotic hand! (The rest of the robot it outside of our view.)
The smaller thing could be a human, too. Giant, good looking but creepy child holding small vulnerable human in one hand, looking at it emotionlessly. But MIRI will not like this version, because they really want to avoid anthropomorphizing the AI.
A Darwinian image of a speck, a mouse, a monkey, a human, and a computer. Optionally, several computers, each exponentially larger with the last going completely off the edge of the page (i.e. building sized). Optionally, with the brains highlighted in some sort of x-ray kind of view.
Edit: Nevermind, I see this is a listed cliche. Well, it’s cliche for a reason.
The concept I think would important/effective to convey is the scope of the vast capability gap between the human mind and a sufficiently advanced AI.
The image that comes to my mind is (a slide from one of EY’s talks) of a continuum showing the village idiot & Einstein to the left as cognitively near-equals, and AI waaaaaaay over to the right toward the I’m-so-smart-it-is-scary zone.
A dark close-up of a security camera. In its lens you can see a threatened-looking human.
A chess board, human pieces vs robotic. The human is in checkmate.
An opened laptop on top of a pile of skulls.
A robotic hand grasping the world. See 1, 2
A depressed-looking scientist putting his stuff (books, papers, funny gizmos) in a cardboard box on his desk. Behind him, a huge mainframe computer. In the background, screens showing complicated diagrams, and robot hands manipulating chemistry equipment.
Similar possibility: An observation/control room for a large factory/server farm/ whatever that is completely automated. The overseer is being replaced. This might do well with some visual reference to Valve’s Portal.
An advanced, automated chemistry/physics lab: robot arms and cameras everywhere, particle accelerator, beakers, tubes all over the place, QR codes. Nothing that looks made for humans.
I would vote against any kind of chess metaphor—when I think of chess-based AI I think of faster CPU power giving faster searches of possible solutions, rather than the impression of “true” intelligence (avoiding the definition of what that actually is for now).
Only idea I had was the image you see a lot of developing humans—like the chain of apes to humans, but with the robot at the end. That could be played straight with the robot facing right with the implication of replacing the previous generations, or played more sinisterly with the robot turned back to look at the last human with a menacing or indifferent expression. Alternatively, have several robot generations getting more unrecognisable but include a timeline linked to the previous generations to show the speed new generations are developing. But it’s maybe been overused for humour at this point.
A human looking at an ape with a smirk, in a zoo. (one half) A computer “looking” at a human, with a smirk, at home. (the other half) It would have to be cartoony for the computer to “smirk”
A shot of an abandoned chess board with the king toppled (optional: and a robot hand shaking with a human).
or
A chess board with one grain of rice on the corner, two next to it, then four, (...) until the board and surrounding world is rice.
I’d go for chess or similar since people already associate chess with “computers are better than humans at this game of smarts.” Still, it’s cliched so maybe you want something less so. As for the examples, 1 will look really campy, 2 is uninteresting, 3 is okay, and 4 is good.
I like to point out that a several studies on design fixation indicate that describing solutions that you don’t want but seem relevenat to the problem can be counterproductive. Also, the postive examples coming before
seems like a case of failing to hold off proposing solutions.
We got a “hold off proposing solutions” comment on the “book title brainstorming” post, too. But remember, we discussed the general issues quite a bit internally before proposing solutions; LW just isn’t seeing that part of the planning.
Yes, but the discussion here will be anchored by the solutions you’ve proposed, even if the discussion that created those solutions was not anchored by them.
I think that was the idea? ‘Hey y’all, we want a design that’s kind of like one of these.’
http://www.dailygalaxy.com/.a/6a00d8341bf7f753ef019affc63311970d-800wi
My first thought was that cartoon where a human is listing things only a human can do, with all the things computers can already do falling to the floor.
I tried to search for the image, then found out that I originally saw it in lukeprog’s Intelligence Explosion book. So I guess you know about it.
Still, it’s a great image: http://intelligenceexplosion.com/2011/superstition-in-retreat/
A humanoid robot in a lab coat, building an even more impressive-looking robot (but less human-like), in a workshop with scattered high-tech stuff. On a whiteboard, advanced-looking designs and diagrams.
or
An evolution of different kinds of writing:
chalk writing and glyphs on a wall
elaborate pen writing (Shakespeare or something)
A kind of brainstormingish diagram on a whiteboard, written with colored pens and doodly drawings
a dense and complex diagram, that doesn’t look produced by a human (QR codes instead of writing, very abstract
or
A high-tech futuristic city, but without anything human-related or human-understandable (cables and pipes and QR codes yes, windows and writing and posters no). On the ground, a few scattered human bones.
Of a dark, scfi-fi blue hue:
Extending from the left towards the centre of the cover is a translucent plastic computer screen we view from behind, through which we see complicated code displayed in white text. Beyond the screen is arena seating, reminiscent of the UN, with world leaders’s heads fallen dead at their desks, beside their country’s flag-emblazoned name plate.
No, the problem of uFAI is not that we will have to surrender to it.
It is that the uFAI will destroy us as a side effect of achieving its goals, when it uses our resources for its purposes.
You’re right. I was assuming we might parlay for subjugation; you made me realise that whatever marginal benefit our assistance would confer the uFAI, the marginal chance of our destroying the uFAI precludes enslaved coexistence.
Could somebody create a poll for the above options?
I guess Luke wants us to create new options, not vote on the existing ones.
This said, here are my feelings from looking at the pictures:
#2 and #3 are “someone is playing with the font trying to impress me”, which repels me for similar reasons articles using many font colors repel me; it feels desperate
#1 feels like an elementary-school textbook
#4 feels like a novel
So I think #1 would be great for a gentle introduction into the topic of artificial intelligence for lay audience; and #4 would be great for a book like GEB. But I guess the prepared ebook is neither.
Hmmm… Here are some that focus on AI risk…
How about some scientists looking at a monitor (which is turned away from the reader) with mixed expressions of wonder, shock, horror, and confusion. Possibly you could have one of the researchers desperately trying to smash the computer or turn it off, while another is desperately trying to hold him back. Sort of trying to capture the failure of a real-life AI boxing…
A little more dramatic: A group of police or soldiers who appear to be arresting a server farm. One of them can see a monitor and looks really, really, panicked, or defeated, or is just facepalming.
A scary-looking Terminator-like cyborg, but dressed in a tuxedo or something, handing a (turned away) paper to a human. Either the human could be really confused or shocked or something, or the cyborg could be making a creepy “I am not left-handed” smile.
Not Exactly a march of progress line, but something like a chimp and einstein on one corner and a server rack on the far end. Similar to the line diagram used in the sequences to illustrate how much of a difference we;re looking at. We are looking at appealing to numerate people, so it should not be overkill to have a graph.