Google’s A.I. now has Artificial Imagination

Google has modified its image recognition neural net with a feedback loop.. and it now “dreams up” images of its own, even from nothing…

Google imagination (photo: Google)

But if the neural network is tasked with finding a more complex feature – such as animals – in an image, it ends up generating a much more disturbing hallucination:

511161f2-d7bf-45b9-81ed-ab391fed3cf9-620x467 (photo: Google)

Ultimately, the software can even run on an image which is nothing more than random noise, generating features that are entirely of its own imagination.

c5fe2bba-8035-4174-98da-ff896075b8c5-1020x680

“One way to visualise what goes on is to turn the network upside down and ask it to enhance an input image in such a way as to elicit a particular interpretation,” they add. “Say you want to know what sort of image would result in ‘banana’. Start with an image full of random noise, then gradually tweak the image towards what the neural net considers a banana.”

(source here)

The image recognition software has already made it into consumer products. Google’s new photo service, Google Photos, features the option to search images with text: entering “dog”, for instance, will pull out every image Google can find which has a dog in it (and occasionally images with other quadrupedal mammals, as well).

So there you have it: Androids don’t just dream of electric sheep; they also dream of mesmerising, multicoloured landscapes.

Source article: http://www.theguardian.com/technology/2015/jun/18/google-image-recognition-neural-network-androids-dream-electric-sheep

Artificial Intelligence wrote 2,7 million WikiPedia articles

For This Author, 10,000 Wikipedia Articles Is a Good Day’s Work

Sverker Johansson

Sverker Johansson


FALUN, Sweden— Sverker Johansson could be the most prolific author you’ve never heard of.

Volunteering his time over the past seven years publishing to Wikipedia, the 53-year-old Swede can take credit for 2.7 million articles, or 8.5% of the entire collection, according to Wikimedia analytics, which measures the site’s traffic. His stats far outpace any other user, the group says.

He has been particularly prolific cataloging obscure animal species, including butterflies and beetles, and is proud of his work highlighting towns in the Philippines. About one-third of his entries are uploaded to the Swedish language version of Wikipedia, and the rest are composed in two versions of Filipino, one of which is his wife’s native tongue.

An administrator holding degrees in linguistics, civil engineering, economics and particle physics, he says he has long been interested in “the origin of things, oh, everything.”

It isn’t uncommon, however, for Wikipedia purists to complain about his method. That is because the bulk of his entries have been created by a computer software program—known as a bot. Critics say bots crowd out the creativity only humans can generate.

Mr. Johansson’s program scrubs databases and other digital sources for information, and then packages it into an article. On a good day, he says his “Lsjbot” creates up to 10,000 new entries.

On Wikipedia, any registered user can create an entry. Mr. Johansson has to find a reliable database, create a template for a given subject and then launch his bot from his computer. The software program searches for information, then publishes it to Wikipedia.

Bots have long been used to author and edit entries on Wikipedia, and, more recently, an increasingly large amount of the site’s new content is written by bots. Their use is regulated by Wikipedia users called the “Bot Approvals Group.”

While Mr. Johansson works to achieve consensus approval for his project, he and his bot-loving peers expect to continue facing resistance. “There is a vocal minority who don’t like it,” he said during a recent speech on his work. Still, he soldiers on.

“I’m doing this to create absolute democracy online,” Mr. Johansson said recently while sitting in front of a computer at his office at Sweden’s Dalarna University.

Wikipedia, he reckons, should someday be able to tell people everything about everything. His bot, which took him months’ worth of programming to create, is a step toward achieving that goal sooner rather than later—even if the entries it creates are bare-boned “stubs” containing basic information.

Achim Raschka is one of the people who would like Mr. Johansson to change course. The 41-year-old German Wikipedia enthusiast can spend days writing an in-depth article about a single type of plant.

“I am against production of bot-generated stubs in general,” he said. He is particularly irked by Mr. Johansson’s Lsjbot, which prizes quantity over quality and is “not helping the readers and users of Wikipedia.”

Mr. Raschka says these items “only contain more or less correct taxonomic information, not what the animal looks like and other important things.”

Others have echoed his concerns on public chat forums, comparing Mr. Johansson to “rambot,” a bot used to add county and city articles in the U.S., that contain only the most rudimentary information.

Source: http://online.wsj.com/articles/for-this-author-10-000-wikipedia-articles-is-a-good-days-work-1405305001