Turing's Test is from 1950. We don't judge dogs only by how human they are. Judging software by a human ideal is like a species bias.
Software is the new System. It errs. Some errors are jokes (witness funny auto-correct). Driver-less cars don't crash like we do. Maybe a few will.
These processes are our partners now (Siri). Whether a singleton evolves rapidly, software evolves continuously, now.
Crocker's Rules
The definitions of words are a pragmatic matter; we choose them to make concepts easy to talk about. If the definition of AI were broadened to cover all software, then we would immediately need a new word for "software which is autonomous and general-purpose in a vaguely human-like way", because that's a thing which people want to talk about.
When you start thinking about "chauvinism", with regard to software that exists now, it's... kind of like if someone were to talk about how people were being mean to granite boulders. I'm just scratching my head about how you came to believe that.
We don't judge dogs only by how human they are
No, but we do judge dogs by how intelligent they are. And there are certain dogs that are more intelligent than others. Intelligence != human intelligence. Furthermore, most software only interacts with other software/hardware/firmware. To the extent that it interacts with meatspace that interaction is mediated by a person. AI would be software that interacts efficiently with meatspace directly without human intervention.
If AI is software is AI, then human intelligence is DNA is human intelligence. An obvious non-sequitur.
Take google maps. On one end, it interacts through road sensors. On the other end, it serves us by telling about traffic.
So, if I pick the piece of software that happens to be closest at hand, which in this case is the browser with which I am reading your post - you claim that Firefox is an AI? It's a complex mechanism, sure, but so is a car, and we don't generally regard those as intelligences.
What do you believe the term AI actually means, if a Hello World program apparently qualifies, but a rubber stamp of the words Hello World does not qualify?
If you look at the way MIRI defines AGI you won't find it mentioning the turing test as the primary criteria.
As far as addressing the issue of the Turing test Bruce Sterling's article http://www.wired.com/2012/06/turing-centenary-speech-new-aesthetic/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%253A+wiredbeyond+%2528Blog+-+Beyond+the+Beyond%252FSterling%2529 is a lot better and a lot more fun.
https://www.evernote.com/shard/s138/sh/e59e45f0-6f5e-df08-e225-4cdf804cd58d/531c325084d8d7f815658e1889c64e8e
Well Andy, you have discovered the buzz saw which is the misnamed (based on Karma results) "Discussion" section of this website. Of course the Accepted Theology around here is that lesswrong.com is NOT a cult, so we need to come up with some other explanation for the sheer undiscussability of certain ideas.
Do explain, mwengler: Are you arguing that we should upvote Andy's claim that bugs are supposedly intentional jokes played on us by a playfully childlike software?
If you believe I'm misconstruing/misinterpreting what Andy has been saying, I'll show you the original text of this post of his, before he edited it.
Do explain, mwengler:
Since you ask, OK.
When I got to this thread, top level was downvoted -30 and many of Andy's comments were downvoted -5 or more. I looked at Andy's comments other than in this thread and many of THEM were downvoted even though they appeared innocuous, and Andy was at -117 (IIRC) for recent downvotes.
Are you arguing that we should upvote Andy's claim that bugs are supposedly intentional jokes played on us by a playfully childlike software?
As an historical matter, I was not arguing that "we" should do anything. I was intending to signal to Andy that he was not unique in experiencing this result when interpreting the name "discussion" as an invitation to, well, discussion. There is no mechanism here to PM such a thing that I am aware of. Although I didn't mind signalling this publicly as long as it would not be too expensive, which is a typical result when going against the crowd here.
If I were to argue something that "we" should do, it would be that there be a cost to downvoting, like there is on the stackexchange family of sites. Downvoting can still be done, but it costs the downvoter a karma point. A rationalist community, founded on a lack of faith in authority, would benefit in my opinion from a higher level of unpopular irreverence than the current system produces. Sure, an echo chamber is valuable, but we will still have plenty of popular mainstream echo posts even if a few irreverent posts are allowed to be discussed.
If I'd gotten to this thread and it was at -4 and the comments inside were mostly at 0 with some give and take between Andy and his detractors, that would seem about right to me. This is discussion. Andy has an informed viewpoint that will resonate with some other people. It makes sense to have a function here where whether or not there is anything to what Andy says can be discussed. Maybe Andy learns something from that. Maybe his detractors learn something. Maybe I learn something from it.
Alternative to allowing discussion posts, maybe this section should be renamed. Bullpen. Sandbox. Auxiliary. StagingArea. Something that would make it clear it is NOT a discussion area.
ArisKatsaris, I did not say "intentional", and if I did, I would not mean it in quite the way you do. But I would mean it in an analogous way to our intentionality. Why did you insert this word?
This is another call for respectful dialog on the topic. Takers?
A brief word on credentials. I am a 23/24-year "veteran" of the software industry. I have worked on many types of software at Microsoft, and on simulation and optimization at Electronic Arts. I am an information scientist first, and an "armchair" theoretical physicist (with a pet TOE), and a hands-on consciousness researcher.
Thank you for the civil dialog.
and an "armchair" theoretical physicist (with a pet TOE)
Publicly admitting this, while brave, results in me and probably others revising the probability of the stuff you post being useful or interesting way down. This is because you don't understand that in physics it takes a decade or so of dedicated studying to reach the proverbial shoulders of the giants, which is necessary before you can figure out anything new. It's the same in math, and probably in many other sciences.
If you want to ask interesting questions, let alone contribute non-trivial insights, start by familiarizing yourself with the subject matter, be it physics, cognitive sciences or AI research.
Seconded. Any time I hear someone has a pet TOE, I dramatically revise my opinion downward - it happened with Wolfram, and now it happens to you. Even the highest end physicists that I'm aware of make no such claims, other than vague statements like "I suspect X is more likely to be correct than Y."
To be fair to Andy, having a "pet TOE" can mean something as simple as feeling one TOE is more "elegant" or whatever. It's not necessarily the case that they are one of those guys who thinks they've single-handedly "solved physics."
This needs be 10s of times longer.
Human style intelligence us the only example we have of human level intelligence.
I agree with you last point. LW has no right to focus all the attention on singeltons.
I am suggesting your comment implied to me you still compare AIs with humans a bit too much. We work to make software able to solve the set of problems it was designed for. This applies for Hello World, and for Singleton.
Single-use software has its place, but it's not exactly singularity-inducing. Each piece of software can only do one thing. If you had a piece of software that could do anything, then you program that one piece of software and you and everyone else is set until the heat death of the universe.
Also, why bother with the word AI? Even if AGI isn't its own cluster in thingspace, we already have the word "software". Why replace it?