Hacker News new | past | comments | ask | show | jobs | submit login
Designing a 'Young Lady’s Illustrated Primer' from Diamond Age (medium.com)
153 points by arthurjj on Apr 24, 2018 | hide | past | web | favorite | 104 comments



I feel like the article is severely overestimating the state of the art and the current capabilities of the "big 5".

All of them are working on the most basic fundamental components that are prerequisites to a Primer and every prerequisite is still pretty far off: no one can currently build a useful conversational agent (anything beyond query-and-response of Alexa), no one can really extract useful semantic meaning from text comments with surrounding context, no one can construct good automatic customized lessons for 5th graders based on feedback loops. There is no meaningful version of a primer that can exist without all of those things just being solved problems.

Given the inability to meaningfully execute in any of those areas, the challenges listed in the article seem laughably meaningless to me: the problem of "they won't use it 24/7, so the system wouldn't be able to learn everything it needs to" is meaningless if we can't make something learn something from 24/7 use. Similarly, a system diagram showing that it would be a book with sensors and smarts doesn't really seem to be providing any insight at all.


I think you're missing the point, which is to develop something that can help people learn better today and be a starting point for something like the primer, not to build the primer today. Obviously the technology for the latter doesn't exist, which is why some creativity is needed in approaching the problem. The state of the art in guided learning is still textbooks and lectures, and most applications of technology beyond those have been failures, but that doesn't mean there are no wins to be had with current technology.


I think it is the exact opposite of that. There is some version that can be done with current technology, but the challenges for that version are fundamentally much much more basic than what is discussed here: the article is basically jumping ahead imagining we had 90% of the technology of the primer solved and what problems you might face in that hypothetical world where the remaining constraints being "people don't talk to it enough, otherwise it would work"

The article mentions several times the concerns about getting enough user specific data and that it can't be easily bootstrapped by a smaller company that isn't the "big 5" because others don't have enough data. Except the big 5 can't solve related and easier problems yet with the data they have and a ton of engineering effort invested. The article ponders whether a primer for the hardest topics (educated adults) could have enough usage while we can't even do "make something that usefully, intelligently adapts to teach users even trivial topics or targeting a single grade of very young children", or "something that can hold a conversation better than a toddler".


I think the author knows and is assuming a lot of what you're pointing out. The technology for the Primer as such isn't there and won't be until we have general AI. However, we could build something today with current tech, like a Kindle that teaches you to read, or teaches you calculus, and it wouldn't be hard to make a compelling product. There are many possible starting points and the technology isn't really the problem if the goal is just to make a start. The real problem is much deeper and the key point is that the Primer was subversive. How do you organize a consumer product company around a subversive product? You don't. So how do you fund such a project? These are much bigger than the technical problems.

There's a reason why Stephenson had the Primer commissioned by an infinitely rich eccentric and built by a rebel engineer. What I suspect the author is hinting at by talking about data isn't that the data is hard to get, it is that for any company in our society with access to that level of resources to start making machines to mold young minds would be a pretty short road to a pretty boring dystopia.


This is close to the inspiration for the One Laptop Per Child project, as far as I am aware. YLIP was a major philosophical impetus for it, and it was even named "Nell" after the main character in Diamond Age.

The Amazon Kindle was also codenamed "Fiona" after another main character, Fiona Hackworth.


> This is close to the inspiration for the One Laptop Per Child project, as far as I am aware. YLIP was a major philosophical impetus for it...

The OLPC was actually Nicholas Negroponte's second attempt at getting inexpensive computing to the developed world; I know because I worked for his first attempt at the CMIRH in 1982 & 1983. The Diamond Age wasn't even published until 1995.

This isn't to say that many OLPC folks weren't inspired by the book, but the OLPC's inspiration goes much further back.


Is the OLPC considered a successful endeavor?


In an act of great timing, there was just a pretty decent article about this on The Verge [1].

[1]https://www.theverge.com/2018/4/16/17233946/olpcs-100-laptop...


In itself, probably not. (though I can't speak for the children and teachers who have gotten to use them in the real world). However, it changed the landscape for computing - netbooks, iPads and Chromebooks all arguably descend from olpc.


That's probably generous. OLPC may have helped create an aspiration for cheaper portable computers. But netbooks themselves, as originally conceived, never amounted to much either. iPads are mostly consumption devices. And Chromebooks, while I'm a personal fan, are a marginal success.

Tech advances have made it possible to buy pretty inexpensive laptops but I'm not sure how much credit OLPC deserves for that.


OLPC showed that it was possible; before that companies didn't seem to be interested in seeing if they could. True netbooks never amounted to much, but they did show that there existed a market for basically a consumption device (web browser and movie-watcher). Which the iPad took up and make much more usable.


I love this book. Still very current despite the fact it was published in the nineties. I re-read most of Neal Stephenson's stuff every few years. Just finished Anathem, again ;-).


I like Stephenson's writing a lot but i don't know if I could re-read his books. The frequent, lengthy digressions put me off of reading them again.


IMO Diamond Age feels way less like a recap of Stephenson's reading list than some of the others. Worth a second look if you're on the fence and looking for a good story.


It's always felt like the most disciplined of his books to me - Snow Crash is almost there, but just a little too formulaic as a story. I enjoy the digressions and length in his other work, but they can get a little dull after the first reading.


I agree as far as the novel/ridiculously-long-essay dichotomy goes. That said Snow Crash is usually called a satire of (post)cyberpunk and as a satire it needs to adhere to the formula more than just another entry in the genre.


That explains a lot about my impressions of Snow Crash specifically. I felt that Stephenson wasn't being serious with his writing (contrast to, say, William Gibson's original treatment of cyberpunk). A bit toward the Terry Pratchett side of the writing spectrum.

I found it to be a slight turn-off, but if what you say is true, I'm more inclined to give Stephenson's other work more attention.


I wonder how that lines up with people who hate the first chapter and think it’s a nonsequitur versus people who enjoy it for its absurdity, versus people who think it is perfectly in keeping with the rest of the book.

Is it Gene Wilder doing a front flip in his first scene in Willie Wonka so you never quite trust him again?


>> Gene Wilder doing a front flip in his first scene in Willie Wonka so you never quite trust him again?

...

...

Oh SHIT.


You are one of today’s 10,000. Go read up on Gene Wilder and all the feedback he gave for WW. That scene was his idea. He got into that role like Daniel Day Lewis or Viggo Mortensen.


Naming your main character Hiro Protagonist gives the game away somewhat.


Unless you're 12 years old and Snow Crash is your first exposure to the genre.

<.<

>.>

You know, speaking for a friend.


Even so, your ‘friend’ could have chosen worse places to start...


To be perfectly honest, I did (and still do) oscillate on how I see that choice. Is it:

1) A clever if superficial name,

2) A dumb name because it's too obvious,

3) A clever name because it's satirizing the dumb choice of an obvious name,

4) A dumb name because the satire is also transparent...

Basically it only works on odd "Yomi layers" (see http://www.sirlin.net/ptw-book/7-spies-of-the-mind).

The rest of the book perfectly straddles the "haha-only-serious" line so I guess it's in keeping.


> usually called a satire

Isn't that a post-rationalisation? Was anyone saying that at the time?


I don't know. I first read it in like '98/'99 ish? and by the time I was reading reviews for it people were calling it satire or at least definitely winking at the audience (as a sibling points out: hiro protagonist) I don't know what the conversation was contemporaneously with release.


Oh yeah. I read it when it came out and it was definitely satire. The genre's tropes were very well set up by then, actually now that I think about it, it was probably the gravestone for the genre.


I think the point of the prologue in Diamond Age is to act as gravestone.

Bud (Nell's father) is the prototypical cyberpunk character. And he's literally killed off by the events of that chapter. He plays no role in Nell's life, he's never brought up again, it's Stephenson saying this novel is literally post-cyberpunk.


What is the relationship of the Young Lady's Primer to Colin, the souped-up, hand-held travel guide used by the Yakuza boss's little girl in Mona Lisa Overdrive? Satire or just hyperbolic derivative?


I actually liked it less than others.

Snow Crash, despite being a train-wreck in terms of narrative cohesion, plotting, and basically every other measure of fiction as a craft, is just about perfect mind-candy. I reread it about two years ago, and it was almost as good as the first time.

Later work like Anathem, I think, starts showing him as a mature author[1], finally paying as much attention to the craft of writing as the ideas.

Diamond Age... I dunno. Certainly had its moments, and the exploration of 90s-era anarcho-capitalist crypto-anarchy themes was solid, but it really didn't grab me that much. The writing was less uneven than Snow Crash, but a lot of the crazy brilliance was missing.

[1] Still flunking Fictional Narrative Endings 101, but apparently that's intentional.


[1] Still flunking Fictional Narrative Endings 101, but apparently that's intentional.

How is that intentional, I’m curious? I love his books but I hate hate hate HATE how he can’t ever finish one properly. I just went through seveneves and decided that maybe I really had enough of his endings.


He used to have a comment on it at one of his websites, but that appears to be gone. This quotes that: http://kaedrin.com/weblog/archive/001495.html


His endings have consistently left me expecting a sequel. Well, not really expecting, any more, because I pretty much know what (not) to expect. But then there was Mongoliad, so ???


One of the reasons I love his books is the lack of "Standard Hollywood / Fictional Ending number 1" that has become so pervasive that it's effectively satire of itself and so unsatisfying.


Those are frequently my favorite parts. When I think of Cryptonomicon, I think about the Cap'N Crunch problem.


I’m currently re-reading Seveneves, and will probably tuck into Cryptonomicon (for the third time) in the next few months. Neal Stephenson’s books are rich enough that I usually get a different vibe from them when I read them again. For instance, on first read I liked the second part of Seveneves better than the first part, and now it’s the other way around. Maybe it’s due to the fact that on first read I get sucked in by the plot twists and on later reads I can enjoy the scenery and pick up different themes.


Seveneves was deeply traumatizing to me; I don't think Stephenson really grasps what 5000 years means to some isolated populations.

I don't think I'm ever going to re-read it.


Yeah, I was/am a huge fan of his books. Anathem is just an incredible work of art and deserves every bit of recognition it got if not more.

But REAMDE and Seveneves were just terribly disappointing and a huge letdown to me. They both seemed shallow and flat compared to the depth of story in his other books.

I don't think I can think of any other author whose books I both adore and loathe.


REAMDE, among other problems, was one of those books where all sorts of things had to just click together or there wouldn't be a story.

Seveneves was pretty good for about half the book. Then illogical events happened that culminated in "and then a miracle (really multiple miracles) happened" which really stretched my suspension of disbelief for the last part.


A couple of days ago I've decided to not finish Seveneves after having been disappointed by REAMDE already. Are there any recent book releases anyone can recommend to people who love Cryptonomicon, Anathem and Snow Crash but aren't really into REAMDE and Seveneves?


It's not necessarily that Seveneves felt 'flat' so much as the second half presents people as basically un changed after millennia which in some cases is over-lookable and in other cases deeply deeply lacking in empathy.


I admit, I also felt the characterization of people in the second half was upsetting, but more upsetting to me was the enormous loss of diversity, and I thought that somewhat explained the relations. These people aren't humans as we know them, they are walking scientific caricatures of them come to life. From that, I didn't really feel it was a statement about "people" per-se, as much as it was a way for him to manipulate the situation in a way where he could make wild predictions about millennia in the future while skirting reality fairly loosely. All quirks and weirdness can basically be explained as "see situation".


As extreme a genetic bottle neck as he invents I still feel like his future is overly distinguished. But as I say in siblings: thats not what wrecks me.


But hadn’t the pingers started to develop adaptations to ocean life? And both the diggers and the pingers would have a hard time generating novel adaptations because they were severely population constrained (smaller population = lower total generation of mutations). And anyway, 5000 years is enough time for subpopulations to diverge, but it’s not enough time for subspecies to emerge.


They would have likely (almost definitely since it was a government project) had access to similar gene editing tech to the Ark.


The spacers adherence to the past struck me as un realistic. The Diggers though I can't even think of their millennia, the absolute iron rule they had to maintain to survive... the inescapable fear... well I'm crying as I type this comment.


> The spacers adherence to the past struck me as un realistic.

I don't know. If you are the last remnants of a great civilization, I can see that being elevated to the point of religion or near religion, which can help things persist for quite a long time, since it's longevity is no longer based on logic.


...and it boils down to a glorified horoscope. Dinans are impulsive Teklans are reliable etc etc. sigh.


I look at it this way, each member of that race grew up in a culture that practically deifies the Eves and the choices they made wrt disposition and traits, given that the traits are partially drilled into each generation as they grow up. Toss a bit of (epi)genetic predetermination and it's pretty easy to see how these groups wind up conforming to stereotypes like that.

Also we're only getting Kathree's view into things too so who's to say how accurate the strict caricatures actually are?


It's not the spacers that make me break into tears a year after I read the book though.

Realistic or not they had thousands of years more freedom than the Diggers.


It's harsh and it's sad, but is it really all that much worse than the many tens of thousands of years where humans lived in small tribes, sometimes in harsh ecosystems, with disease and pain and local conflict?

For the vast majority of our history, humans have lived far worse off than now. Sure, it's easy to tell yourself they were happy living in a simpler time in and some cases a tropical paradise, but losing most your siblings in early childhood and passing away yourself at the old age of 32 because you scraped your leg and tf got infected doesn't sound all that great to me.

That said, people of the past probably were happy. Happiness at your condition and situation seems to be highly relative. To us, it looks like a horrible punishment, but to them, it's just life, and losing your friends and family to illness and accident was likely as accepted as the fact today that you'll at some point die of old age.


Ten thousand years ago they could walk away, they might die but they wouldn't kill everyone that they could conceive of. The resources weren't obviously limited, the idea that everything they could conceive of would be lost wouldn't have occurred. The absolute limitation, the sacrifice, the need to keep sacrificing for un told generations...

THEY HAD NO SKY. They had no future but theory...

I'm not trying to convince you. I'm just saying: it was traumatizing to me.


The races and the inter-racial relations in the book's future make zero sense: Why the hell did Moirans and Teklans not intermarry into a single race centuries ago? Why are the docile sexless Camites still around? Why are there not one but TWO races that are mainly characterized as being evil (and the characterization is not even subverted in-story)?


I don't think I really grasp what 5000 years means either, and to which populations? I liked Seveneves because I'm a big fan of stories set in space, but I'm interested in hearing your take on it if you feel like writing more.


Nothing particularly. The spacers just seemed overly concerned with 5k ago (which for reference is almost pre history for us).

nccebkvzngryl n uhaqerq crbcyr yvivat va n pnir sbe 5000 lrnef gur nofbyhgr veba svfg gb fheivir vg. Gur greebe bs nalguvat snvyvat. Gur furne genhzn bs 200 trarengvbaf tebjvat haqre guvf nofbyhgr hadhrfgvbanoyr yvzvgngvba...

Fgrcurafba unf ab vqrn whfg ubj qrrcyl gung jbhyq qrfgebl hf. Tenagrq gung'f cnegyl sbe gur fnxr bs univat n cybg va gur frpbaq unys, fgvyy... Nggrzcgvat gb rzcunfvmr jvgu gurz qrfgeblf zr.


But we would probably be more concerned with 5k years ago if humanity were highly developed enough to have left us an extensive collection of writing. Heck, Hindus and Jews still care about writings (orally transmitted before the alphabet) that originated 3-4K years ago.


Ooof. Stephenson is among my favorite authors. Before Seveneves that wouldn't have had a qualifier. I re-read Cryptonomicon every year or two, the baroque cycle about half that frequently, and Anathem a couple times now. I own Snow Crash and The Diamond Age in first editions. I cannot imagine re-reading Seveneves; it was both not a totally flawed plot (a quality I can ordinarily forgive in a Stephenson novel -- I love a good yarn) and it killed everyone I know and love, in an immediate near future sense, on the page in front of me.

It was, for lack of a better description, a callous novel.


If I were to re-read Snow Crash, I would skip all the talk about ancient Sumer and the Sumerians. IMHO, that book would be greatly improved by eliminating all of that.


It serves to build and explain the neo-scientologist motivation of the antagonists. And in my opinion it is somewhat more sane than real scientology mythology.


I've only re-read "snow crash" - but that was actually a listen to audiobook on a roadtrip


> TYLIP is...a book that is powered by a computer so advanced it’s almost magical, and it teaches children everything. It does this through a fully interactive story. It teaches you how to read, how to do maths, it teaches you morals, ethics, even self-defence.

http://mssv.net/2006/05/01/the-young-ladys-illustrated-prime...


An interesting take away from the article is that games like Animal Crossing might be a better staring point for building TYLIP than the methods in the OP article


100% agreed. We are taking a game based approach to creating the primer.


"Scaling one-on-one tutoring is seen by many experts and researchers as the silver bullet for human cognitive development"

While I admire the author's intentions, this is completely the wrong approach. There are multiple reasons why having actual human tutors is important:

0. The creation of a deep, long-lasting relationship between tutor and tutee

1. The benefits to the tutee: the feeling of pride in helping the younger generation, building empathy and understanding of what the younger generation is doing

2. The tutee is inspired to pass along their knowledge to others later in life; they learn how to tutor by being tutored, thus continuing the cycle

That last point is important. Even if a perfect Primer were created, how would it improve with time? If it truly replaced tutoring, then the art of tutoring would be lost forever, and it'd be up to software engineers to determine the metrics by which to measure the success of the Primer's tutoring algorithm.

Of course, that would never happen. Instead, tutoring by Primer would be reserved for poor people who didn't have tutors in their lives--or who had potential tutors who were scared off by the notion that they were not as effective the Primer. Then, in those communities, organic tutoring--an institution that has existed since literally the beginning of humanity--would be amputated and replaced by whatever tutoring the Primer's developers saw fit.

Real tutoring would, of course, be preserved for the ruling elite. Which I think is what happens in the book, come to think of it.

Of course I could be completely wrong. But this is the peril of MOOCs in general: students who go to first-tier schools like Harvard and Oxford get personalized instruction, while everyone else gets video lectures while glorified TAs proctor exams.[0]

0. https://www.chronicle.com/article/Why-Professors-at-San-Jose...

'Professors in the philosophy department at San Jose State University are refusing to teach a philosophy course developed by edX, saying they do not want to enable what they see as a push to "replace professors, dismantle departments, and provide a diminished education for students in public universities."'


I haven't read the book in years, but from my recollection, there were three girls given the Primer; the narrator of the primer was a cloud-provided mocap/voice actor, running from an AI generated script.

For one of the girls, a single actress performed the majority of it, for the another it was whomever was available, the third would be a bit of a spoiler, so I won't say here. This was pointed to as one of the differences, with the single "tutor" being superior (their socio-economic statuses were far different as well).

It is true in the book that scaling it for the masses required pure-CG and was considered significantly inferior. Personalized instruction by an expert in the field has always been the realm of the privileged, so it's not necessarily a fault of a technology that it doesn't change this.


It's not clear that the results for Nell are "better", all three girls who recieve Runcible in its original form have outcomes as prescribed. They become independent of the Victorian mainstream. They aren't like the other girls. Nell founds a new group allied to the Victorians, Fiona is last seen among secretive London actors in some unspecified undertaking, and Elizabeth joins CryptNet, whose true purpose is anyone's guess.

The primers given to the Mouse Army are not just Runcible without a ractor. Hackworth reflexively, even unconsciously proposes something more, he says something about them being made culturally appropriate or something. Doctor X doesn't see this for the threat it really is. All those little girls are being set up by Hackworth to depend on an outside leader, whereas the girls who receive Runcible are completely independent. And the Mouse Army answer not to Doctor X or even to Hackworth but to Nell.


I had missed the point that Hackworth's cultural modifications were what led to the mouse army. I'll have to read it yet another time.


> 0. The creation of a deep, long-lasting relationship between tutor and tutee

I have spent over a decade in school and this has literally never happened to me. I went to second rate middle/high schools, so maybe teachers in "good" schools have some deep insight which makes interactions with them actually desirable?

In any way, this desperately does not scale, at the current ratios you would require a single teacher having "deep, long-lasting" relationships with ~hundred students at once, which just isn't feasible.

Even if every adult in the society got involved, the proportion drops by just about an order of magnitude. I can't imagine myself tutoring a single child, let alone ten of them!


Tutoring children are what parents (are suppose to) do. Maybe if society enters the much talked about post-scarcity mode, humans tutoring each other is something we could spend a lot more time doing. I know I'd enjoy it.


But why would we spend precious time personally tutoring our children when we could use the earnings we would forfeit to bid up real estate in neighborhoods with good schools, where our children can compete for 1/30th of the attention of somebody who probably doesn’t care about them all that much?


I believe the key word is: "scaling"

Great tutors are great. Irreplaceable. There are approximately 900 of them in the world. Good luck hiring one of them for your child.

For an example of what kids can do in highschool, when well tutored - by excellent private tutors, look up: https://en.wikipedia.org/wiki/Vitalik_Buterin

AYLIP could be the major contribution of the recent boom in AI. Not just in enriching a few scummy FANGs, but actually providing real value to society.


Where do you get the 900 figure from?


You're missing the point entirely.

Yes, human tutors are better, but what do you do when human tutors aren't available? Or when they can't teach you something you want to know?

That's exactly the story in The Diamond Age, the Young Lady's Illustrated Primer lands in the hands of a girl with no money and an abusive father.


Not abusive, dead. Bud dies in the prologue. The guy sexually abusing Nell is another of Tequila's boyfriends, and her brother's violent street gang "solves" that problem.


While I agree with your overall sentiment, a primer could improve over time if it were a true AI. Though that would be concerning in itself as then at some point, the AI decides what humanity's children should learn, and can start to manipulate human society. Though maybe it would do a better job than we have been.


Why shouldnt the same thing as happens in the book happen on a voluntary basis? A second app connecting to the tutor, basically providing open source education..


Diamond Age included a remote human tutor who acted as various characters in the imaginary world of the YLIP.


I'm a little troubled that this seems to be completely missing the point of the Diamond Age (aka, "the Fight Club effect"). The whole lesson of the Young Lady's Illustrated Primer is that you shouldn't make the Young Lady's Illustrated Primer.


Strange that you are getting downvoted. The second half of Diamond Age is all about how Hackworth’s manipulated version of the book, given to the Chinese orphan girls, causes them to grow up into educated and skilled robots who need a strong external leader in the form of Nell. Nell’s success comes from the fact that she was lucky enough to also have strong role models.

This should strike even more alarm bells at a time where we are worrying about the pernicious effects of social media and maliciously targeted material (including material targeted to children).


I upvoted them. My question was legitimate, because I'd forgotten about this part:

> the Chinese orphan girls, causes them to grow up into educated and skilled robots who need a strong external leader in the form of Nell.

They're capable, skilled individuals with no individuality. At once a positive (work well together) and a negative (seemingly require external leadership to direct their efforts). So a downside to the mass adoption of the book.


I saw the book as relatively neutral on this point. Nell certainly is improved by it (though it could be easily read that this is largely due to Miranda), and Elizabeth's malaise is easily explained by growing up in luxury, with an abusive teacher.

The YLIP is shown as something inferior to a true parent (Miranda being Nell's surrogate parent being pointed to as a big reason for her success), but better than nothing (it was a "best availabe" tool Dr X. provided for the orphans).


Wasn't Elizabeth's tutor-actor different from day-to-day? (If I'm recalling my characters correctly.) And that was partly the explanation for how she turned out. While Nell's was a dedicated tutor-actor through the whole time.


This is true, and more support for the the book considering the Primers to not be insufficient replacement for an involved parent, which is an easy argument to make, but does not mean the book thinks they are bad.


Nell's tutor started recognizing the job offers for the tutor lessons and choose to become the dedicated one.


I'm curious how you got to that conclusion. I'll admit, it's been ~16 years since I read that book. I remember most of the story but I didn't have the same reaction you seem to have. I'd like to know what your reasoning is.


Indeed.

Honestly, this is all human work. Rather than try to take all the value out of doing it, we really should embrace doing it, and fund those willing to do the work.

Sure, not all of them will be stellar, but some will be. That's worth a lot.

Seems to me the biggest benefit tech has brought us is so much information access that many only need guidance, experience, to be challenged, mentored.

I give back to young people in this way every chance I get. Right now, it's hard, but it's not always hard. Most of those people have come back to say thanks. It helped. They are on a path, chasing after it, whatever it is for them.

Good as it gets right now, IMHO.


An HN story about Croquet (Smalltalk framework) dropped a few years ago. Someone was like "So, Alan Kay is trying to build the Metaverse?" My response was, no, Alan Kay is trying to build the Young Lady's Illustrated Primer from The Diamond Age, and always has been. The Metaverse is one of the lemmas he has to develop in order to get there.


And Alan Kay would also cite John McCarthy's Advice Taker as a reference for such a project, which this article didn't do. https://en.wikipedia.org/wiki/Advice_taker


"based on ... (mumble) ...builds on ... learner's existing knowledge " (https://proto-knowledge.blogspot.com/2011/11/building-young-...) is a key part of any such system. A hard part, since it requires building a map of the student's 'database'. Quizzing without being too intrusive. And secured.

(Easier than in a classroom. Typically faced with 2-dozen or more students, the teacher has little idea what curriculum each student has been exposed to. Assessing the group wrongly means having to go back and start all over. And the only possible solution still has to be general, not tailored.)

This has been and remains a wonderful dream application. If it's for all, not just for the privileged, bravo.


For mature learners, a prototype they might find useful would be a research assistant. The AI could work at making sure it understood queries better over time, then get better at finding answers, then get better at summarizing why its findings are useful, and finally work on explaining/teaching what it found.


this is more like "the Librarian".


I've been working on this for a year or so (unrelated to this article). I assume the AI will be available at some point in the future. The hard part is actually the teaching material.


Seems to me that among other things the writer has completely missed the point of the Primer was not to educate in the sense they mean it but to build character.


From a post above is

> I feel like the article is severely overestimating the state of the art and the current capabilities of the "big 5".

To me this is especially diplomatic!

From the OP

> Obviously we have the Primer,

WHAT?????? What do "we have"? Where is this "Primer" that "we have"?????

What here is "obvious"?

IMHO this might as well been written in Russian, and there I don't even know the alphabet!

For decades we've had multiple choice tests. Is THAT what she's talking about that "we have"?

So, could write a computer program that asks multiple choice questions and for wrong answers displays some background tutorial material. Is THAT what she has in mind?

She keeps saying "AI", that is, "artificial intelligence": Just what more specifically does she have in mind? Curve fitting with networks of sigmoid functions, with regression trees, with versions of regression?


So much of education when a child is young is made up of non-academic cognitive skills. How do you shape and react to a child's human interactions - and the associated cognitive skills - via any kind of media or device? It's hard enough for trained human beings...


In The Diamond Age, it was very advanced AI that included software specifically aimed at social interactions -- plus a kind of trust fund that paid for hiring "racters" (character actors) in certain circumstances.


I always though "racters" was short for "interactors", but it could be either one...


I think it's a bit of both the 'shows'/programs are called "'ractives" clearly for interactives, from there we can get ractors from interactors or just a perversion of actors in ractives.


Oh, I think that might be right! It's been a little while. :-)


I haven't read the book. It monitor interactions with other people or just the device?


From what I remember (couple of years since I read it) it can keep track of other people if they are near by but mostly the child would talk to the book like a friend and explain what was happening.


Wow, for a mature mind :), that was a lot to absorb.

It’s interesting to see an overview of the current state of art. Still, its too bad the conclusion is that only the big 5 have a chance (Google, Facebook, ...)


Why limit it to a 'book'? I feel that the people behind the movie Her has hit it and it's the design that everyone is trying to make real.


One of those books i simply noped out of after the n-th time it diverged into lala-land.

Stephenson may know his technical stuff, but it does in no way make him able to write interesting books.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: