You are using an older browser version. Please use a supported version for the best MSN experience.

Meet Articoolo, the robot writer with content for brains

TechCrunch TechCrunch 28/06/2016 Natasha Lomas

There is great writing. And there is content. And at this juncture in the Internet’s evolution it seems very plain, Dear Online Reader, that you are mostly being served a tsunami of content — accelerated into your attention trough by click-dependent digital business models which require a steady stream of word fodder to engage eyeballs long enough to ambush them with ads.

You can weep for the state of the written word — and let’s face it, a whole lot of content isn’t even written these days; I mean why expend energy composing actual sentences when you can just livestream a melon being slow-ploded by rubber bands? — or you can build technology to automate the writing process to churn out even more #content.

The second option is what Israeli startup Articoolo has been busy doing. It’s built an algorithm that can generate an article on a topic of your choice — so long as it can be described in between two and five words. Just sum up your subject concisely, tell the machine how many words (max 500) to scribe on your behalf, and select whether you prefer “better readability” or “enhanced uniqueness”. Then hit the create button and sit back and wait for your article to be baked by its AI.

Assuming, that is, your topic is not too obscure for the algorithm to tackle. (In which case it will reject your request and ask you to reword / pick from a list of similar topics to further drain the strain from the creative process.)

When there is enough source material in the robot’s content databanks it will get to work, progressing through its automated research and drafting process. The time taken to deliver your content will vary, apparently depending on article length and topic complexity (and perhaps demand on Articoolo’s servers). The speediest article I managed to generate was a superficial sample of less than 250 words on the topic of ‘Kim Kardashian’ — which took the algorithm less than 50 seconds to churn out. Although the result was more a spot treatment on cellulite than a comprehensive appraisal of Kardashian’s career.

Other subjects definitely broke the minute-to-create mark. Asking the robot to serve up advice on ‘How to be Happy’ seemed, oddly enough, to take it the longest of all. To be sure of this, I asked for content on this exact same topic twice — and got the same article, slow baked, word for word, twice. So initially I thought there must be some pretty hard limits on the robot’s capacity for original thought. But Articoolo claimed this was a bug; a passing glitch in its infinite copy generating matrix. And lo and behold a third attempt on the topic delivered a fresh stream of robotic agony aunting…

“Don’t dwell on negative ideas, kick them out and gradually encourage yourself instead. Give yourself permission to grin or laugh out loud as you recall. Do something to make another person feel good, it’ll brighten your day overly.”

“The algorithm will write you a different article every time,” confirms CEO and co-founder Doron Tal. “Even if you ask for the same topic 100 times. You will get 100 different articles. We actually have a feature for affiliate marketing companies that enables users to create multiple articles on the same topic at once.”

The team launched their online copywriting service a few weeks ago in beta, after some two years dev work on the core algorithms — aiming to test the water to see how much bona fide commercial demand there is for automated content generation. Of course there is already plenty of content being churned out online by humans to keep other humans clicking. But how much of a leap is it to flip the switch to autopilot and pump out purely algorithmic arranged thought?

Tal, a self professed “digital advertising expert”, says the team founded the business at the end of 2014, after the two other co-founders had been playing around rewriting Wikipedia articles for laughs — which in turn sparked the idea to automate copywriting for ad-driven purposes after he got involved and, he says, had his wicked way with the concept. They’ve been seed funded by three local VCs to tune of $1.2 million at this stage.

As you might expect, the initial target customers for Articoolo’s service are SEO and content marketing types. So it’s not out to replace yours truly just yet. Although the startup claims the results generated by its robot writer are both original and readable. And their pitch email to me was the attention-grabbing one liner: ‘will journalists lose their job soon?’ [sic]

After testing out the service, the truth of the matter is rather less clearly and pithily framed (although the grammar mistakes were at least algorithmically sustained).

Less clearly framed unless, that is, you’re convinced your ‘reader’ is unlikely to be doing much more than skimming lightly over the surface of your content like a stone hopping atop a glassy lake, tripping from one hackneyed thought to another, eyeballing words as if they’re so many pretty shapes.

Fans of cryptic crosswords, who relish the mental adrenaline kicks they get from being forced to slalom from cliche to climax, with precious little in the way of linkage, may also be unlikely appreciators of the robot oracle’s semantic gaps.

Normal, average readers, however, will struggle to locate much in the way of sense.

And discerning readers will read this content and weep. The robot’s output is about as intellectually fulfilling as a backwash of used tea leaves.

Learning how to live a happy life’s about learning how you work. Whenever you make the choice, the next thing is to set clear targets of what you need to reach in each among the primary areas of your life,” opines the robot scribe, once again on the aforementioned subject of ‘How to be Happy’, mining its eternal font of received wisdom.

Here is the thing: Trust your very own intuition as to how you would like to use your goals. Do you need to review them every week? Do you need to rewrite them monthly? life’s about going with the flow, but also to find out how you perceive the world determines your well-being.”

Cliches, tautologies, non sequiturs. This robot might actually be better employed writing speeches for politicians. Although its bland pronouncements feel like they need to be spiced up with emergency! exclamation! marks! after! each! and! every! word!

Each one of these stunning sights have to be seen to be enjoyed thoroughly,” the machine pens on another well trodden topic, getting its grammar in a twist again. “The Seven Wonders of the World possess some good magic to them. You can visit them and live an unforgettable experience for life” it drones on with wonderous banality hammered home by additional tautology — right before switching gears and dropping this unexpected last liner: “Jimmy Singh is connected with Wonders of The All World“.

Jimmy Singh? Who he?

“From the foot spa to the countless designer watches there appears to be nothing you can purchase her that she doesn’t already have. Never underestimate the power of the box of hand made chocolates or perhaps a designer silver teapot. A coupon for something she’s never done, but want to attempt is always an excellent present idea. This could be something from salsa dancing lessons to sky diving. You can find actually countless experiences you can buy coupons for and web sites that list them all in one spot. All you should do so is choose the right one and you will be giving her a distinctive and thrilling gift.”

If you’re wondering what input topic elicited this deadpan parade of stereotypical and strangely unexciting suggestions for a lady who apparently has everything yet still desires handmade and bespoke curios, or at least a coupon to go skydiving, it was a request for an article about ‘chocolate teapots’.  Go figure.

“We still have work to do on the quality,” Tal admits. “We wanted to find out if there’s a need for a solution like ours and if target audience is ready to pay for it. Our overall strategy is to provide solution to various industries where content is a significant building block.”

‘Building blocks of words’ reminds me of the infamous phrase ‘binders of women’ for some reason. I wonder why that is?

“No, we don’t think our algorithm will be able to replace human journalists, at least in the near future,” he adds. “Algorithm doesn’t have a point of view, an opinion or an agenda. It can be a helpful tool for reporters. You can count on the algorithm to do most of the background research and provide the base of your article.”

Personally I’d view a set of tea leaves/coffee beans, properly brewed in hot water, as a far more useful tool for aiding and abetting the article creation process — but call me an old fashioned journalist…

But okay, okay, personal preference aside, the ugly reality is that content farms continue filling the Internet with tasteless fodder. And these entities’ appetite to cement more words into ever more numerous clickable units is only going to grow — at least until all human communication has been rendered down to livestreamed slow-ploded melons to keep our eyes perpetually peeled.

So if you consider rapacious appetite for ad padding ‘a problem’ then Articuloo is offering ‘a solution’ of sorts — albeit probably just another gap filler, treading water until digital marketing missives can be wrapped unmissably across people’s eyeballs via some kind of vision disrupting augmented reality.

We’ll all be Clockwork Oranges then.

In the meantime, far from helping with the wider problem of the Internet’s abusive relationship with the written word, Articoolo is doing its best to accelerate the slope of the onramp that feeds intellectually vacant slabs of content online like a donut factory feeds a conveyor belt. Clickbait written by robots does feel like we’re approaching rock bottom/peak content.

Tal says the startup has kicked off pilots with several companies in the native advertising, affiliate marketing and ecommerce spaces at this stage. Although he is not naming names, nor quantifying customer sizes as yet.

The current beta version of the service asks users to buy credits to pay per article created (or rewritten — a secondary feature that’s supposed to enhance your article’s SEO potential). But the aim, should they be able to prove out enough demand for robotically generated content blocks, is to move to a SaaS model, feeding the digital void with a steady stream of machine curated ‘word textures’ — which I think is a phrase that more properly describes the quality of the commercial output at this germinal stage.

How exactly does their algorithm turn a blank page into a wall of text? The various steps involved include: analyzing/understanding topic context; parsing a specific set of related resources (using an indexed source database of ~500GB of “trusted articles”, and if it can’t find what it’s looking for there resorting to the open web); extracting sentiment and “important keywords” from the source material in order to turn out “one coherent piece of text”; rewriting this text, using an NLP engine, for “multi-level semantic identification and to verify its readability”, as Tal puts it.

“The most difficult part is the rephrasing process,” he adds. “We actually had to teach our algorithm how to use English correctly. It is very complexed and we keep on doing that in order to improve the results.”

So human grammar nazis can chalk up another advantage over the robot.

Talking up the plus sides of the word machine, Tal points out it’s faster than a human writer and can also scale to levels of productivity that not even the sweatiest and least sociable blogger could hope to achieve.

“No human will be able to write thousands of articles per day!” is his triumphant claim.

Albeit, even at 50 seconds a pop (for the shortest, simplest articles) the robot writer is still not going to be able to turn out 2,000+ articles in 24 hours. More than a thousand, sure. But not multiple thousands yet.

In any case, whether anything except a machine would willingly digest the thousands of pieces of content that could be conjured into existence by Articoolo is a whole other question.

But then it’s highly probably that the vast majority of these algorithmic content works will languish unread. Because in the ever accelerating game of ‘click here to eyeball this’ the battle to slay human attention spans has been fought and won. Economics has defeated sense. Call it a sort of reverse Brexit if you will. Progress, apparently, marches in many directions these days.

I’ll leave you with the robot writer’s passing thesis on quantum physics — which at times appears to bend towards a warning on the fickle fluctuations of new media — as a specimen to feed into a wider debate about the worth of algorithmically generated content. And indeed the general proliferation of #content online.

In the words of the bot: “In this chaotic time between the change of one Age into the changing of the brand new age thought and belief systems are running rampant… Don’t believe anything you hear and only ten percent of what you see.”

Amen to that at least.

The free encyclopedia Wikipedia, tells us that Quantum Entanglement is just a physical system where two or more things are connected with each other even when they’re spatially distinguished. It talks about spin measurement and hidden variants, Bell’s inequity, H. Study regarding the smallest particle of issue – the physical Quantum. This is where the Quanta is explained as one soul, the smallest neighborhood of the infinite soul, Head of God. There is no real reality, except as detected by a Quanta – an individual spirit. No individual people, nothing blinks out from the Quantum Ocean, Head of God, no physical reality.

Let’s realize that all is spirit, energy, non real shaking. It is true that every single soul blinked out from the Quantum Ocean, Head of God to experience Life on the real plane. It is also true that the ideas that this personal soul chooses to believe may draw energy out of the ocean Quantum, Mind of God into his\/her aura. The energies one carries in one atmosphere attracts one’s physical reality. What needs to be better comprehended is spiritual Quantum Entanglement. Which implies that even though a person soul attracts, it’s physical life by ideas and beliefs, they’re also influenced by the ideas and beliefs of others.

A belief is merely a thought that has been approved and believed about over and over. On a conscious level a soul can be trying very difficult to pick and select the ideas they think and also to believe the beliefs which they believe in. In this chaotic time between the change of one Age into the changing of the brand new age thought and belief systems are running rampant. You can go outside on the Net and purchase a dozen new belief systems to get a dollar. Or they’re being electronically spewed outside at us from the mass media.

Each time you turn the Television on or read a paper, magazine, see a bill board you’re being bombarded by another person’s thoughts. These thoughts and impressions might not necessarily affect or harm you do not believe them. Do you believe everything you hear or see? That is a classic definition of present day insanity. To remain sane in these difficult transitional times, we need to embrace the idea: don’t believe anything you hear and only ten percent of what you see. It’d be more helpful for mankind if the Quantum Scientists would study Quantum Soul Entanglement. Teach the people more about ideas and beliefs transfer between living souls.

More from TechCrunch

image beaconimage beaconimage beacon