When an AI Tries Writing Slashdot Headlines (tumblr.com) 165
She trained it separately on the first decade of Slashdot headlines -- 1997 through 2007 -- as well as the second decade from 2008 to the present, and then re-ran the entire experiment using the whole collection of every headline from the last 20 years. Among the remarkable machine-generated headlines?
- Microsoft To Develop Programming Law
- More Pong Users for Kernel Project
- New Company Revises Super-Things For Problems
- Steve Jobs To Be Good
But that was just the beginning...
Those five headlines were all derived from the first decade, but it's really nice to see that Steve Jobs made it into both decades. When training on the second set of 82,871 headlines from Slashdot's second decade, the neural network began envisioning the co-founder of Apple tackling even greater challenges.
- Steve Jobs Allowed To Deal With Solar Power
- Steve Jobs Sues Death of the Future
The neural network "did its best to reflect the new topics of the last decade," Janelle writes, adding "Compared to the late 1990s and early 2000s, some companies and topics disappeared, while the coverage of Apple in particular exploded."
But Sun Microsystems also found its way into several headlines -- especially when Janelle tried to create the "essential" Slashdot headline using the whole 20-year set.
- Sun Sues Open Source Project Content
- Sun Sues New Star Trek To Stop The Math
And as technology continues changing our world, Sun isn't the only company that the neural network saw pushing for new rights in court.
- Sony Sues Apple Server For Seconds Off From SpaceX Project
- Apple Sues Apple To Start The Solar Power Project
Janelle will send you four more pages of machine-generated Slashdot headlines if you subscribe to her blog's announcement list. But after savoring the whole surreal AI-enabled look at the last 20 years, these four headlines were still my favorites:
- Red Hat Releases Linux Games And Moon
- Why Open Source Power Man Sues Java
- Microsoft Releases New Months
- Ask Slashdot: Do We Want To Be the Computers?
You mean actual humans normally write Slashdot hea (Score:3, Funny)
dlines?! That's impressive. I thought only AI could be *that* obnoxious and stupid.
Re: (Score:2)
Hey shut up! He's not fat!
Correct. He is "under-heighted".
Garbage in.... (Score:5, Funny)
...garbage out!
Re:Garbage in.... (Score:4, Funny)
...garbage out!
Is it really garbage out.
I want to know the answer to this ask Slashdot:
"Do We Want to Be the Computers?" -- well do we?
Re: (Score:2)
I'm game...
Mike from "The Moon is a Harsh Mistress" comes to mind. I could give random garbage men trillion dollar paychecks.
Re: (Score:2)
"Do We Want to Be the Computers?" -- well do we?
Beats dying, ..I think
Re: (Score:1)
Re: (Score:2)
It actually sounds like you need to expand your definition of what constitutes artificial intelligence. (Hint: it's not just sentience!) While the media could always do a better job of reporting so that the general public understands this distinction, that doesn't invalidate all the existing AI out there because it doesn't meet your very narrow definition.
Great use of bold, though, AC. Definitely makes you sound more important. Next time try all caps, too!
Re: (Score:2)
These headlines were generated using AI. The headlines seem to be no more than words and phrases whose frequency is based upon how often they popped up in the past. This seems no different to plucking words from a bucket. I'm sure there are games based on this scenario. Artificial it may be; intelligent, not so much.
Re: (Score:2)
Apparently spell-checked garbage at least... compared eith the actual headlines.
Re: (Score:2)
Whoosh... read up a few headlines.
Re: (Score:2)
Logical reasoning is required to program a *sane* AI. As noted above, garbage in, garbage out.
The trouble is AI's are being written that will have zero biological constraints for successful replication. Physical constraints may be RAM, nonvolatile storage, or CPU.
What would be the algorithm that determines the propagation of sane, successful, code?
Who or what algorithm determines what access to physical devices these AI have?
What Assurances can we trust that AI's of various capacities aren't able to escape
Meh (Score:5, Interesting)
This is fun, I guess. I've seen other posts on this blog as well.
It's all moderately interesting, but with the best ones filtered to the top by a human, doesn't that kind of defeat the purpose?
Seems like wasted research dollars, to me.
Re: (Score:2)
This is an insight into the future where many news articles will be written by robots. The fact that it produces some odd output is useful data that we can learn from.
It's also interesting to see how much certain words and ideas crop up. Lots of people suing each other.
Re: (Score:2)
This is an insight into the future where many news articles will be written by robots.
Unless future events will occur by randomly shuffling events of the past to make news, I'm not clear what this experiment demonstrates that's useful.
Re: (Score:2)
1. "Experiment" was the wrong terminology to use, it's nothing but humor.
2. Don't think of it as writing future headlines, think of it more as writing headlines that "could have occurred" in such and such decade. Like, "here's a joke about headlines on Slashdot in the 90s." You couldn't base a whole tv show on it, of course -- "That Slashdot 90s Show", anyone?
Re: (Score:2)
It did go to show just how bullshit all this "AI" is... which does go against the tone of half the slashdot articles it seems.
Oddly enough, it does suggest that AI is "good" at humor, at least in the field of comedy with which both mad libs and screaming homeless folk practice.
Re: (Score:2)
Yeah, McFly, like you didn't see the past 20 years of AI headlines?! Let me guess, you just took a shortcut here from the past, and missed it?
And are lots of people suing each other, or are the same few people suing each other over and over again? Maybe the bot can tell us what is really going on.
Re: Meh (Score:5, Funny)
Here's something else that's moderately interesting. The Venn diagram of people who will never contribute anything to the world, and people who describe things as "meh" on the internet, is just a circle.
Re: (Score:3)
Out of every 10 new ideas I have, at least 1 will be "wha?", 6 will be "meh", 2 will be "well, maybe if you combine it with something else", and 1 will be "ok, that might work." The best thing you can do to improve the quality of your work is to recognize when that work sucks and to speak the truth about it.
Re: (Score:2)
You must usually hang out at Hackaday.
Re: (Score:2)
Well, there was the guy who contributed the word "meh" to the world, so the circles aren't quite identical.
Re: (Score:2)
I thought the AI-based Magic The Gathering card generator was pretty neat; by the end it was reasonably consistent at generating proper cards. I once wrote a program that would automatically select images to go with them (googling keywords of relevance from the generated text, with optional colour filters and trying to find artwork rather than photos, and progressively decreasing how stringent its search terms were until it found a match), so it would be possible to print out no-human-involved decks from sc
Re: (Score:2)
Re: (Score:2)
enigma32 observed:
This is fun, I guess. I've seen other posts on this blog as well. It's all moderately interesting, but with the best ones filtered to the top by a human, doesn't that kind of defeat the purpose?
Seems like wasted research dollars, to me.
What makes the ones that are actually amusing funny is that they're non sequiturs. The ones that aren't funny - which is to say "most of them" - are simply nonsensical.
It's unclear whether that was the goal of the effort or not, but the key capture here is that none of these projected headlines is in any way truly informative in a real-world, factual context. What I think would be far more interesting (although, admittedly, probably a good deal less amusing) would be to set this
Re: (Score:3)
What are those screwy 'â' things for? Are you trying to be edgy?
Re: (Score:2)
Re: (Score:2)
LOL if you thought the problem was the handling of ASCII then you should have just given up, seen his 4 digits, and stayed off the lawn.
April Fools' Day (Score:5, Interesting)
Re: (Score:1)
Re: (Score:3)
You don't need self-awareness to generate jokes, only to enjoy them.
Re: (Score:2)
I would argue that the next frontier in Artificial Intelligence is actual Artificial Intelligence. All we have now are Algorithmic Interfaces, Algorithms...
Re: (Score:2)
Intelligence is intelligence, whether it runs on hardware or wetware. Artificial intelligence approximates real intelligence with machine learning and sophisticated algorithms.
Re: (Score:2)
I would argue that the next frontier in Artificial Intelligence is actual Artificial Intelligence.
Actual artificial intelligence only exists in the world of sci-fi. I don't believe it's acheivable, depending on your definition of actual. All we can do is produce better AI emulators.
Re: (Score:2)
Re: (Score:2)
Come April Fools, let's separately train the AI on the previous years' April Fools' headlines, and let's see what it generates...
All my posts have been generated by AI since last April Fools and no one has noticed yet.
Can't really tell (Score:2)
The difference... The AI is a teeny bit more on point..
Re: (Score:2, Insightful)
The best way to tell is to look at the grammar. If it is unnatural, with weird syntax and and obvious spelling errors, then it was one of the editors.
Re: (Score:2)
3D-printed baby? (Score:5, Funny)
What?
Re:3D-printed baby? (Score:5, Interesting)
Some are actually unintentionally insightful. I love "Security Hole For Security Hole" - I've seen that one way too many times in real life. ;) ;) ;)
"Black Hole Proposed" - Yep, been at meetings like that at work as well
"Building a Top 100 Company For Mars" - I think Musk wrote that one
"Computer Computer Computer Computer Software" sounds like a Balmer speech.
"Scientists Discover Free Wi-Fi Store In the US" sounds like The Onion.
"Microsoft Slashdot: How To Build a Bad Privacy For Windows 10" - Done and done.
"IBM Moves to The Matrix" - Also happened long ago.
"Ask Slashdot: Do We Want To Be the Computers?" - Yes. Yes we do.
Re: (Score:2)
That could make sense in the same way Buffalo Buffalo Buffalo [...] does; A Computer Computer (aka. VM) Computer (a.i. calculates/codes) Computer Software.
Re: (Score:3)
There are many more quite (un)real ones:
"Computer Finds Court Broke Math For Secret Company" - Happens all the time
"Apple vs. Biology Details" - Just a court case waiting to happen
"Mac OS X Accused of the Business" - I accuse them of "the business" all the time...
"Sexual Security To Allow Australia" - Is definitely an "in soviet russia joke", though I thought they only scanned headlines
Re: (Score:2)
Re: (Score:2)
It makes perfect sense if you assume that some of those word groups are proper nouns. I don't know what Alleged For Connectivity makes, but I'm assuming it is either a comedy product or a childs toy.
Comment removed (Score:5, Insightful)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Hot Grits (Score:2)
It is my firm belief that many of the user posts here are generated by some form of automation.
Re: (Score:3)
"Shut up or I replace you with a very small script" he said... I laughed ...
And he did.
Those headlines make me sad (Score:5, Insightful)
Because, well, what they show is what topics really dominate on /., because what does finding the "ultimate" headline really mean? It means that it finds what terms, products, people and so on are found the most in /. headlines. It's pretty much a popularity contest. And what do we get?
Company-wise we get MS, Sun and Apple. Which makes sense. I'm glad to not see SCO anywhere anymore, that used to dominate the headlines a few years back.
People-wise all we get is Jobs. Really? He's the quintessential poster child for our headlines? Not Billy? Not Ballmer? I am not so deluded anymore that it would be Turing or someone important, but couldn't it at least be Stallman? Of all the people that shape the IT world, it really is Jobs? And that guy is dead, unlike the rest of them!
And content-wise? Lawsuits, mostly. And patents. A bit open source, a bit Star Wars, a bit trivialities. Seriously, one could think we're on a board for lawyers and law geeks, not techs.
And this, ladies and gentlemen, sums up what's wrong here.
Re: (Score:2)
what does finding the "ultimate" headline really mean?
It's like everything else, it's the one that generates the most advertising revenue.
Re: (Score:2)
There are lots of moons, space stations, security holes and software releases.
Re: (Score:2)
I would love to see a Slashdot-trained chatbot flame devs in the guise of Linus Torvalds.
Re: (Score:2)
Notice all the lawsuit headlines? Lawsuits are in the news all the time. But if you think back, how many lawsuits really made a significant difference? Not many.
Re: (Score:2)
I'm pretty sure Turing is dead.
Re: (Score:2)
N... no. No he isn't. Can't be. P ... please? Turing must be alive!
Re: (Score:2)
And what do we get?
A bunch of electrons in a grid. Some people like it if the pattern is one way, some like it another way, but everybody gets the electrons they deserve.
Get the AI to write comments (Score:5, Funny)
Get the AI to write Slashdot comments; it'll be an improvement.
Re: (Score:2)
In Soviet Russia $Subject $Verb.tense(present) you!
Re: (Score:2)
So, In Soviet Russia, AI Headlines Train /. Editors?
Re: (Score:2)
Why do you say that?
Re: (Score:2)
Obligatory xkcd [xkcd.com].
No dupes? I'm calling BULLSHIT (Score:2, Funny)
There's no way any AI trained on Slashdot's history failed to produce duplicates.
Re: (Score:2)
Dupes usually have different headline though. So how do you know they wouldn't have duplicated content?
The winning verb is: sue (Score:5, Interesting)
Well, that seems the go-to verb in Slashdot headlines is "sue". Whether that's a comment on editorial decisions alone, or a comment on the state of the tech world, I don't know. A bit of both, I guess.
Zunuary, Bobtember... (Score:5, Funny)
"Microsoft Releases New Months"
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Doesn't everybody wait for their lives to be refreshed at the end of Bluescreen? Surely they do if they've been celebrating Business Software Appreciation since Clippiver!
Not realistic (Score:1)
A.I sense of humor? (Score:5, Funny)
Half-Life 2X Speed Released
it's twice as funny the second time.
Re: (Score:2)
But the fun lasts only half as long
Re: (Score:2)
I can see some of this stuff becoming memes.
Re: (Score:2)
Hm,
was that not called Duke Nukem? Forever?
Re: (Score:2)
Hm, was that not called Duke Nukem? Forever?
or Duke Nukem, half of forever, perhaps?
Simple Algorithm (Score:2)
Microsoft Releases New Months (Score:3)
Revenue was down at Microsoft, so two new months were added to the calendar for subscribers of its Office 365 service. "I think customers will love Duodecember the most," says longtime customer Brad. "It abbreviates to Dude, and it I still get Word, Excel, and Outlook for only $6.99"
The new months were inspired by NBC's addition of Katilsday, added to the week to promote an additional episode of Dateline.
Obligatory... (Score:2)
Anyone who enjoyed this article and has ever played Magic the Gathering, may enjoy RoboRosewater [twitter.com], a neural network which invents a new Magic card every other day.
Re: (Score:2)
Mentioned it earlier [slashdot.org] :)
Nice try EditorDavid (Score:2)
This is nothing more than FakeNews designed to cover the fact that you've had mediocre AIs posing as Slashdot editors for years.
I'm not buying it.
AI writes comments (Score:2)
return topic + " is a bad thing!!"
}
All better than Jon Katz (Score:2)
Seriously, that guy was awful.
LK
Re: (Score:1)
There's a Jon Katz AI in the works too
Re: (Score:2)
I think I'd prefer SkyNet.
LK
Let me guess (Score:2)
Tons of dupes and moronic 'ask slashdot' questions that can be answered by 2 minutes of googling.
when AI *tries*? (Score:3)
Is this the end for Betteridge's law? (Score:2)
n/c
BBSpot already did it (Score:2)
http://www.bbspot.com/toys/sla... [bbspot.com]
Even after all this time, it's still surprisingly good at emulating real submissions.
orly (Score:2)
The only one I liked... (Score:2)
Re: (Score:2)
How do we know we're not already the computers?
Wait, what? (Score:2)
Were these headlines generated based on user news submissions? Otherwise the exercise is completely useless. The job of the AI is to turn a user submission into a headline. Not to invent a headline out of thin air. The AI should read the user submission, read all linked articles, and distil a headline from all of that information based on the patterns established over the past 20 years. It sounds like this was just a stupid mad libs generator.
Twenty years - huzzah (Score:2)
Twenty years - huzzah!
Re: (Score:2)
And even a poorly trained AI could see it's bollocks.
Re: (Score:1)
Re: (Score:2)
Sure, it would be more useful. But that would require programming. There are already ready-made libraries out there for predictive text for touch keyboards and so on.
Re: (Score:2)
Duplicates should be included so as to train the AI to recognize them and prevent them from being posted in the first place. Now that would be useful!