"Leather rain sheets dug in Harry's ghost as he walked along the land toward the castle, and Ron stands there and makes a kind of dance frenzy, saw Harry and immediately start eating the family of Ermioni."
"Ron Ron's shirt was just as bad as Ron himself"
No, this is not from a dead new Harry Potter sequel – created by Botnik Studios, a creative collection of about 30 writers, artists and developers from around the world who use tools to blend and raise language.
Coachella line ups, Harry Potter lyrics, lyrics to songs … these are just some of the things Botnik Studios has been confused by turning them into strange, viral works.
Prior to the meeting of Botnik's TNW responses, we talked with co-founder Jamie Brew about how the collective company creates such a strange and wonderful job.
How Botnik works
The team started with Google DeepMind wanted to make a bot for The New YorkerHe was created by the author of the magazine from 1997 to 2017, Bob Mankoff. Observing a recovery in similar projects that were trying to "solve" creative tasks with automation, Mankoff and Brew created Botnik, which tried to increase rather than automate.
"We use machines to enable new kinds of human creativity. Botnik Studios … was developed by our users' testers for our first web application, a text prediction keyboard called Voicebox and word suggestions based on any source text that you feed it.
Since then, Botnik has recreated one The boy meets the world episode, tourist reviews of Parisian landmarks, ASMR and Tinder bios video scenarios, including.
Works usually start with an observation – a writer will notice "a kind or a trophy that has a particular way of saying the words he wants to play with." This idea is shifted to the rest of the team via Slack. Then there are various questions, such as how to divide the text in order to use the language most effectively:
"Are we using a neural network that best suits the creation of text fragments that are a couple or a few words We train a prediction keypad throughout the text at the same time We create multiple prediction keyboards that are trained on a subset of text such as when it is separated by characters for our scripts or in dialogue and narration for book chapters? "
Then language tools are created based on the answers to these questions, which are used to write a few lines of text. This happens when man-machine cooperation begins:
"With the prediction keyboard, it's usually a dozen or so writers who put on an editor who assembles them into a Google document or script platform showrunner.io. Each author proceeds through the suggestions suggested by the source text, while the author keeps the writers up to date with the kinds of lines we need today – things like "A Scully Emergency Order" or "A Narrative Line Defining the Scene" . "
"The farthest pieces emerge from the chaos from the bottom up of the semi-random lines and from the guided top-down guidance of the author who is thinking of presenting it in whatever form we are going to."
The process may initially be driven by the machine, but the final results are equally shaped by human hands. "How is it collaborative? Extremely collaborative. It's far from the charts, "Brew says.
Not all project ideas are implemented, usually because the original text is impossible to detect or does not exist: "I would love to train a bot for all the little children your dentist has ever done while you sit there with your mouth open, unable to answer. "
The dentist discusses among other things, says Brew, must to be a 100% person: "No one has collected this data and I do not think anyone will do or should, and that is why the transition to the dentist is so important." You hear the wisdom that Google does not have idea. "
However, human involvement was a two-legged sword. When asked if Botnik received some criticism from people who had been deceived by their work, for example, someone who tried to buy a ticket to the fake Coachella series, Brew replied:I believe the biggest confusion (and maybe the anger) that we have seen is not from people who believe that our work is from a person, but from people who want to read our letter as fully automated by a robot without human participation . "
"This is understandable because there is a full range of art (including writing) in this field at this time, ranging from a truly empirical process generation where the creator acts more like an experimental scientist – by methodically adjusting the parameters and observing the results – acts like an artist with creative control, who can clean production or direct it or bypass it. "
However, Botnik's plans need a human touch.
Botnik recently successfully completed a crowcounting Kickstarter campaign for one The Life, a pop-full-length album, remixing the lyrics from "Scottish folk ballads, Amazon reviews, Carrie Underwood, style elements, and much more, "all about the original music." The project reached 116% of funding and is due to be released in mid-2019.
Currently, the collective team is working on a holiday holiday calendar for early promoters of Kickstarter campaign, "who will be able to call a phone line every day in December to hear robo-messages optimized to deliver seasonal gaiety. "
"As always, we are also working on several secret projects that require the UN's highest security level to know." None of us are working for the United Nations, so none of us knows what these projects are, we follow the rules and send the exit in a graded email address. "
Anything designed for the future? "We'll sort out the food by building a robot that eats your food for you. Subscribe to our feed on Twitter and a newsletter to learn how. "
Jamie Brew and others from the Botnik team endorse us tomorrow on Wednesday, November 7, to host a TNW Answers session. Ask them now NOW!
Everything you have to vote for in the 2018 mid-term elections today