Assisted Writing
Writing is a ancient art form. From Papyrus to MS-Word, the tools we use to write have defined how and what we write. Today, assisted writing tools are using machine learning techniques to understand, manipulate and generate human language. The implications are profound.
This experiment tries to re-imagine word-processing software. It explores new forms of writing, that allow authors to shift their focus from creation to curation, and write more joyfully.
Writing Text
Humans have been writing text, assisted by machines for a very long time. From the first commercial typewriters in the 1860's to today’s laptops, the act of writing has changed dramatically - yet essentially stayed the same: We use our fingers and a keyboard to translate our imagination into matter. The human does the creative work, the machine prints words.
In the 1960s, researchers such as Douglas Engelbart and Ted Nelson started to experiment with ideas such as digital writing tools, the mouse and hypertext. A central goal was to find out how computer systems could augment the human ability to read, write and think. These experiments laid the foundation for a new generation of writing tools.
With the advent of personal computing, digital writing tools went mainstream and became the PCs first killer-app: Word Processors. Still using typewriter metaphors, software such as Nelson’s JOT (1972), Xerox Parc’s Bravo (1974), Apple’s MacWrite (1984) and Microsoft’s Word (1989) quickly solidified the contemporary concept of digital text editing. As MS-Word™ conquered the world, how and what we write changed for ever.
“The medium is the message” -Marshall McLuhan
Re-imagining Word Processing
For all the advanced features digital word processors have given authors, current tools do not address many fundamental questions: How do i write good story? How do i find inspiration & overcome writers-block? In these areas, writing has stayed the same for aeons: The human does the creative work, the machine prints words.
Today, new techniques are redefining our relationship with our writing tools. Since the early days of Auto-Correct, computer assisted writing has evolved into a active research field. Academia and Industry have embraced a range of Machine Learning and Natural Language Processing techniques, to understand, manipulate and generate human language.
The implications of these highly accessible capabilities for word processors design are profound. It is becoming possible to imagine a very different kind of writing software: a tool that actively inspires and guides the writing process, allows authors to widen the frame of reference, shift their focus from creation to curation, and write more joyfully.
Get in touch here: twitter.com/samim | http://samim.io
10 Assisted Word Processing Capabilities
The following capabilities are building blocks for designing new types of word processing software. Each capability references a research project or open-source project to illustrate it’s concept. While the result quality of some projects is still not production ready, it is a useful indicator of where things are going. The shared goal is to mobilize the intelligence of authors and enable new playful writing experiences, assisted by machines.
01. Recommend
Recommend addresses the problem of context. It recommends related texts (books etc.) on the fly, based on the text the author is currently writing.
02. Summarise
Summarise addresses the problem of text length and reading times. It automatically summarises text using neural networks.
03. Simplify
Simplify addresses the problem of text complexity. It automatically simplifies sentences using neural networks.
04. Morph
Morph takes two (or more) text-passages and blends in-between them. A neural net fills in the blanks, generating sentences from a continuous space.
05. Transfer
Transfer addresses the problem of content and style. Thanks to Neural Nets, we can translate our texts into the style of any famous writer.
06. Predict
Predict addresses the problem of flow. It let’s us write much faster. Think of it as a predictive text interface, or “Autocomplete”, powered by a neural net.
07. Generate
Generate addresses the problem of writer’s block. It uses machine learning to generate original texts. Topic & Style are selectable (poetry, lyrics, etc..)
08. Translate
Translate addresses the problem of modality. By translating between media types we can, for example generate stories from jumbled images.
09. Collaborate
Collaborate addresses the pain of writing, alone or in teams. It puts a crowd inside our word processors and let’s us easily crowdsourcing complex work.
10. Interface
Interface addresses the problem of input output. Good speech recognition and text-to-speech system that can master hundreds of accents are here.
And more…
The presented capabilities only scratch the surface of what is currently possible. Re-imagining word processors with machine learning is a very large canvas, transcending traditional notions of reading and writing. One can easily imagine new forms of “writing”, such as auto e-book generation or auto re-enactment by virtual actors of the text one has just written.
Final Thoughts
By combining the presented capabilities, we arrive at a very different view of what a word processors can be: a tool that actively inspires and guides the writing process, allows authors to widen the frame of reference, shift their focus from creation to curation, and write more joyfully.
The coming generation of intelligent content read-writers (word-processors) will be able to understand, manipulate and generate human language in many new, seemingly magical ways. Such writing tools will mobilize the intelligence of authors and enable them to playfully write the next great human story, assisted by machines.
“Writing is easy. All you have to do is cross out the wrong words.” ― Mark Twain
