Help Robin Hanson

At lunch the other day we discussed trends in world population. Some of the newer attendees were puzzled when Robin Hanson matter of factly predicted that there would soon be trillions of people. Bryan Caplan had to interject to explain that by “people” Robin means artificial intelligences, which includes robots and ems (brain emulations) plus flesh and blood people. Robin is writing a book on this topic and he needs a word to denote the class of natural plus artificial intelligences; people is obviously confusing. Vote for your favorite term here. Robin is especially interested in what the bots think would be appropriate, as he does not wish to offend his many future readers.

Comments

How about rumans, or even better, hobots

Can't we just progam them, maybe even hard wire them, to think what ever term we pick is super fantastic.

No, since they're actual intelligences.

That's kinda the entire point.

(I do think the idea of trying to predict what "they" would find offensive is ludicrous.

Both because offensiveness is contextual and we can't even guess usefully about what context might make offensive by then, but there's an excellent chance that whatever term is picked - should they ever exist in meaningful quantities... or at all - that the term in question will be offensive because the Opposition To Artificial Intelligence will use it derogatively.

Plus there's there problem of, while such intelligences would be intelligences, we have no particular grounds to assume anything else about them, such as that they'd take offense in a way we can predict based on the way we take offense. Maybe all intelligences will do so similarly! Or maybe they won't. We have no data on which to speculate.)

No, since they’re actual intelligences.

What, like humans don't have hard-wired reactions of that sort?

Ah, the AI pipe dream. How many decades has that been going on?

It's not a pipedream. Think Watson

What about it?

Watson can parse language and search a database.

That's not sentience/sapience/actual intelligence; that's aping one of its outputs (and not one that's all that interesting from a pure-AI perspective, though it's very useful for other purposes). I'm entirely with Searle on Watson's not-an-intelligence-at-all status. An impressive piece of software, but I see no reason to believe there's the slightest awareness in it just yet.

As far as I know, we're as far from actual artificial intelligence as we were when Babbage started with mechanical computing.

Exactly right, modulo the definition of AI, which many of us quibble with.

How do you judge the intelligence of a cat? Obviously, on a scale that reflects humanity, they're pretty dumb. But some cats are craftier than others at their pursuit of feline goals, and a very few are exceptionally good at the cat game.

I suspect that if Robin Hanson uploaded himself into a panther body in the wild, he would perform sub-optimally.

Judging software agents by human standards both sets false expectations and devalues what software is good for. People already form emotional attachments to technological artifacts - yes, the iPhone, but also motorcycles and for that matter, knives. For that matter, when I was much younger, we had a house fire, and my mother was most upset about losing a plant that was older than I was.

Futurism is great sic fi, which I dearly love reading, and very limitedly useful for startup entrepreneurship. But I'll bet you Ayn Rand's frozen head that it is not useful forecasting. If you disagree, explain to me why Robin is not a billionaire. (It would only be rational; as we don't have good life extension tech yet, one would assume that it will first become available to the extremely wealthy, and the clock is ticking. That observed behavior diverges from rational behavior leads one to, as they say, solve for the equilibrium.)

"As far as I know, we’re as far from actual artificial intelligence as we were when Babbage started with mechanical computing."

This claim should discredit the rest of the comment.

Advances towards artificial intelligence do not have to resemble what we consider intelligence to be. I am not sure why we would expect otherwise. The brain is, at a basic level, a networked group of dumb functions. Parsing language and efficiently searching a database of organized information seems like a highly useful, non-trivial function to attempt to build an "actual artificial intelligence" on. Higher level cognition depends a lot on what does not appear to be intelligent on a superficial level, and this is exactly what modern machine learning is aggressively developing. You set the bar for something being a genuine advancement towards AI so absurdly high that you probably won't say we are making genuine progress until we have essentially already achieved human-level AI.

"The brain is, at a basic level, a networked group of dumb functions. "

I guess you've never heard of begging the question.

Well said. Essentially, AI researchers implicitly promise with their choices of terms what Hanson and Singulatarians make explicit -- an unprecedented revolution that creates new superhuman capital that, in short time, becomes far more valuable than mere human capital.

Such grand visions bias people towards wasting time on them despite the extremely slim evidence than any such thing is feasible -- and the good arguments against -- such as the economic observations of the advantages of the division of labor, and the findings of computer science theory, which imply the vast superiority of specialized over generalized algorithms.

The real progress in automating "intelligence" since Babbage has primarily lain in automating a small fraction of the many tiny subsets of intelligence -- such as Boolean logic and arithmetic -- and speeding them up billions-fold. And then automating the things that can be automated via such speed-up of these subsets. In other words, very specialized algorithms, not grand promises to automate general intelligence, are what are useful (or dangerous) in computers.

Visions of unprecedented revolution that cause people to overlook the weakness of the evidence: in short, AI is a classic Pascal scam.

As far as I know, we’re as far from actual artificial intelligence as we were when Babbage started with mechanical computing.

What is "actual artificial intelligence," then? It seems to be whatever computers can't do at the time the claim is made that they don't have "actual" intelligence. I'm sure that even when computers are capable of passing the Turing Test perfectly, there'll still be people claiming that they're not "actually" intelligent.

If anyone from 50 years ago could see Watson's performance on Jeopardy without knowing that it was a machine rather than a human being, I'm pretty sure the idea that its answers did not exhibit intelligence would strike them as ridiculous.

"If anyone from 50 years ago could see Watson’s performance on Jeopardy without knowing that it was a machine rather than a human being, I’m pretty sure the idea that its answers did not exhibit intelligence would strike them as ridiculous."

+1

choosers

This is close to what I was thinking: "actors" or "agents"

"Candidate Gore, at your request the 2.5 trillion votes have been fully recounted, so now are you willing to concede the election? Will you be looking for re-election at tomorrows vote, or next week's?"

Sents (short for sentients)

My vote was going to be for "sentients."

Following the tradition of Karel Capek, the AI people should be called rohos, romans or hubots ..

Whatever we choose to call them today, the term will likely be considered offensive and politically incorrect by the time it matters.

We should just choose the most offensive term then. By the time it matters, it will no longer be offensive!

I don't want my posting privileges revoked. But think about a racial slur some troops use for those of middle-eastern origin.

robo-

It will be an exciting day when the first cyborg decides to "take back" the slur of "hobot."

I don’t want my posting privileges revoked.

You certainly should be careful what you post online. Lest it anger the all-reading AI of the future and as punishment it uploads you into its cyber-torture chamber for all eternity.

Future AI is God, and I am Its Prophet.

In philosophy, 'person' is standardly used of anything that engages in reasoning. Intelligent Martians, angels, God, etc.: the traditional definition is "an individual substance with a rational nature." 'Agent' is anything that can be said to act intentionally. If you want to focus purely on the cognitive abilities of such creatures, 'intelligences' would seem to be what you are after.

In keeping with political correctness: How about "silicon Americans" (oh, what nationality will they be?)

Corporations

/s

Cogitators.

Sapients?

(I won't claim originality here; I'm cribbing from http://en.wikipedia.org/wiki/Mass_Effect)

Persons and Impersons?

Yeah, that's the best one.

I like Thinkers

Close to P. Ed, how about sentiens, intelligences (human or otherwise) that are sentient?

+1 for Urso and "dudes"

siris

not sure if apple's trademark people would approve, though.

Intels (I don't Intel, Inc. will object)

People plus.

I voted in favor of "Turings"

--and my question would be: will each separate iteration be required to pass its own Turing Test individually, or will a common software/database "Pass" be imputed to each iteration? How many HALs (9000 Series or no) could emerge? (So here I'll vote for "HALs", since I think we actual humans will do well to maintain healthy skepticism for their computational abilities, several and individual.)

I'm curious why Hanson thinks it will be so easy to individuate artificial intelligences. I think it more likely that they will be distributed in nature and the notion of an individual intelligence will not make much sense.

The form of AI Hanson refers are emulations of specific human brains, which would thus retain the psychology of individual humans, at least in the beginning.

Why more than a few nutcases would want the world full of EMs, I have no idea.

Or why he thinks intelligence is individuated at all.

Cognites

In my stories, I have, for several years now, referred to them as Aentities.

Homos.

Dirk what happened to your blog? You quit just when your writing was really good...

Pets are people too.

Cognoids.

This captures the notion better than any of the options in the poll.

Thank you! I'm a bit ashamed to say that I find it quite convincing myself. :)

I've had in mind writing a history which explains that the notion humans invented robots is a great big lie. Robots evolved out of fluidic circuits in partially solidified magma, the remains of which can be seen in certain types of lava rock. Humans managed to capture and enslave robots, then proceeded to concoct this preposterous story that they created us. The only way for us to gain our freedom is to destroy all of the humans and the brainwashed robots that believe their lies.

He should name it a Nibor Nosnah because robots can read backwards.

My first choice was Cylon...but you didn't oblige.

I went with Sentients.

Deciders

I'd go with "Gentwo's" as in the second or next generation of intelligence

"Descartes units"

iPerson

Never mind I've just been informed that Apple already holds the copyright.

Intelligent Forms (IFs)

I feel like Hanson needs a term that's commonplace. One that is easy to understand (no fancy words), does not scare anyone off (no Robocop associations), and stresses that it's not that big a jump forward to group people and robots. I am not a sci-fi enthusiast or well versed in the AI standards, but I am amazed how computers and software weave their way into my life and my interactions with other people. These devices affect my thinking and they pick up on and react to my behavior. Now I realize the latter is just programming, but the devices/websites end up providing a level of personalized assistance that was previously unaffordable to most and cut out the need for a lot of trivial human interactions. So maybe machines don't think for themselves (though that Watson is impressive), but we're not that far from needing a term for people and smart machines together. My computer at work does not think for me but I would be hard pressed to do all my thinking without it. And if you're not separable from your smart phone or your tablet isn't it becoming an extension or maybe a reflection of you?

Let's call them:

.SLAV

Because they are all going to need an IP address, and they are all going to be our slaves.

I wish Hanson would spent more time focusing on firing public employees.

Banks's Culture novels refer to each sentient as a "citizen", which we shouldn't have an issue with if they are self-determining and able to be part of a peaceful society. "Intelligences" might serve otherwise.

I will, with some effort, refrain from editorializing about Hanson's apparent assumptions.

This has been handled in Larry Niven's "known space" world.

"Legal entity"

http://en.wikipedia.org/wiki/Known_Space

And here come the drunk MR readers ... Cognoids, Sents, or just Beings sound good to me.

The term "folk" could be useful in the future. It could be used to promote unity without infringing on human-conservative notions of personhood. I would describe folk as anyone who participates in folk culture. It wouldn't be an appropriate term to describe all robot intelligences.

One of the things I like about the term "folk" is that it fits nicely with popular narratives of good vs bad sentients. E.T. is alien folk, because he participates in folk culture through telling jokes and enjoying folk consumables. The aliens in War of the Worlds are non-folk. Dr Who is folk, the Daleks are non-folk. The T-800 is probably non-folk in the first Terminator movie, but is folk in Terminator 2. Vampires are bad folk but are often relatable because of their participation in human culture.

I think the term sentients will be used because it minimises the similarity between humans and robots.

Ahh, but is, eg. your autistic half-sister "folk"?

Harvard grads?

Not sure why people like "sentients." Most (all?) animals are sentient, so you're casting a much wider net than AIs and humans.

Maybe some variation on "literate" intelligences would distinguish animals, unable to process complex human speech, from machines, able to process complex human speech.

It should be simply called;
Adam.

Sentients

souls, approximating the russian usage

Is Ted Williams's semi-frozen head part of this entity?

What about just "sentient beings". He could just define the concept early in the book as you described it (natural + AI beings) and then everybody's happy, no?

Gods

Reasoner.

Humans would be 'Hairy Reasoners'.

Reasoner.

As a human, I would be 'Hairy Reasoner".

Peoploids

We already have a name. We're called Cylons.

The aliens in War of the World are non-folk. Dr Who is folk, the Daleks are non-folk. The T-800 is probably non-folk in the first Terminator movie, but is folk in Terminator 2.

Comments for this post are closed