The virtuosity of language models like GPT-3 opens a new world of possibility for human-AI collaboration in writing. In this paper, we present a framework in which generative language models are conceptualized as multiverse generators. This framework also applies to human imagination and is core to how we read and write fiction. We call for exploration into this commonality through new forms of interfaces which allow humans to couple their imagination to AI to write, explore, and understand non-linear fiction. We discuss the early insights we have gained from actively pursuing this approach by developing and testing a novel multiversal GPT-3-assisted writing interface.
1. Introduction we have learned over several months of testing and designing interfaces for writing with the aid of GPT-3, beginning by introducing the framework of language models as multiverse generators.
GPT-3 [
We will present the still-evolving approach As the sampling method can be stochastic, running this process multiple times on J1o3i-n1t7,P2r0o2ce1e,dCionlglesgoefSthtaetiAoCn,MUSIUAI 2021 Workshops, April the same input will yield diverging continua" moire@knc.ai (L. Reynolds); kyle@knc.ai (K. tions. Instead of creating a single linear conMcDonell) tinuation, these continuations can be kept and each continued themselves. This yields a branching structure, which we will call a multiverse, or the “subtree” downstream of a © 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 InterCPWrEooUrckReshdoinpgs IhStpN:/c1e6u1r3-w-0s.o7r3g (nCCatEiEoUnUaRlR(C-CWBYSW4..o0o).rrgk)shop Proceedings prompt as shown in Figure 1. lem is not merely epistemic; the future truly has not yet been written, except in probabili2.1. Analogy to Everettian ties. However, when we do finally venture to quantum physics measure it, the ambiguous future seems to us to become a concrete, singular present.
Quantum mechanics tells us that the future is The Everettian or many-worlds interprefundamentally indeterminate. We can calcu- tation of quantum mechanics, which has belate probabilities of future outcomes, but we come increasingly popular among quantum cannot know with certainty what we will ob- physicists in recent years, views the situaserve until we actually measure it. The prob- tion diferently [2]. It claims that we, as observers, live in indeterminacy like the world mension of multiplicity that we must also around us. When we make a measurement, consider, especially when we are talking rather than collapsing the probabilistic world about states defined by natural language. around us into a single present, we join it in Natural language descriptions invariably ambiguity. “We” (in a greater sense than we contain ambiguities. In the case of a narranormally use the word) experience all of the tive, we may say that the natural language possible futures, each in a separate branch of description defines a certain present – but a great multiverse. Other branches quickly it is impossible to describe every variable become decoherent and evolve separately, no that may have an efect on the future. In longer observable or able to influence our any scene there are implicitly many objects subjective slice of the multiverse. present which are not specified but which
This is the universe an autoregressive lan- may conceivably play a role in some future guage model like GPT-3 can generate. From or be entirely absent in another. any given present it creates a functionally The multiverse generated by a language infinite multitude of possible futures, each model downstream of a prompt will conunique and fractally branching. tain outcomes consistent with the ambiguous
David Deutsch, one of the founders of variable taking on separate values which are quantum computing, draws a connection be- mutually inconsistent. tween the concept of a state and its quan- So we define two levels of uncertainty, tum evolution with virtual reality generation which can both be explored by a language [3]. He imagines a theoretical machine which model: simulates environments and models the pos- 1. An uncertainty/multiplicity of present sible responses of all interactions between states, each associated with objects. Deutsch further posits that it will 2. An uncertainty/multiplicity of futures one day be possible to build such a universal consistent with the same "underlying" virtual reality generator, whose repertoire in- present cludes every possible physical environment.
Language models, of course, still fall well We will call the first form of multiplicity short of this dream. But their recent, dra- interpretational multiplicity, and the second matic increase in coherence and fluency al- form dynamic multiplicity. low them to serve as our first approximation of such a virtual reality generator. When given a natural-language description of ob- 3. Human imaginations are jects, they can propagate the multiverse of multiverse generators consequences that result from a vast number of possible interactions.
given a state there are a multiplicity of possible future single-world dynamics; stories unfold diferently in diferent rollouts of an identical initial state. There is another diHumans exist in a constant state of epistemological uncertainty regarding what will happen in the future and even what happened in the past and the state of the present [4]. We are then, by virtue of being adapted to our uncertain environments, natural multiverse reasoners.
David Deutsch also points out that our imaginations, which seek to model the world, mimic reality as virtual reality generators: we model environments and imagine how they could play out in diferent branches.
the multiversal forces at work, the writer’s objective is to eventually compose a single history. However, language models naturally encourage writing explicitly multiversal works.
In the same way that hypertext transcended the limitations the linear order in which physical books are read, exciting a surge of multiversal fiction [5], language models introduce new possibilities for writing nonlinear narratives.
After all, it’s only a small leap from incorporating multiverses in the brainstorming process to including them in the narrative.
Counterfactual branches often occur in traditional fiction in the form of imaginary constructs, and our minds are naturally drawn to their infinite possibility [6].
When a piece of literature is read, the underlying multiverse shapes the reader’s interpretations and expectations. The structure which determines the meaning of a piece as experienced by a reader is not the lineartime story but the implicit, counterfactual past/present/future plexus surrounding each point in the text given by the reader’s projective and interpretive imagination.
More concretely stated, at each moment in a story, there is uncertainty about how dynamics will play out (will the hero think of a way out of their dilemma?) as well as uncertainty about the hidden state of the present (is the mysterious mentor good or evil?).
Each world in the superposition not only exerts an independent efect on the reader’s imagination but interacts with counterfactu- 4. Interfaces als (the hero is aware of the uncertainty of their mentor’s moral alignment, and this in- We propose the creation of new tools to allow lfuences their actions). writers to work alongside language models
The reader simulates the minds of the to explore and be inspired by the multiverses characters and experiences the multiverses already hiding in their writing. evoked by the story. Research into hypertext writing tools has been ongoing for more than two decades and 3.2. Writing as a multiversal act has produced notable tools like StorySpace[7]. However, the issue of hypertext interfaces assisted by language models is a newer development, as only very recently have language models become advanced enough to be useful in the writing process [8]. Likewise, there has been significant research into interactive narratives, including in branching, multiversal settings [9, 10], but never one in which the human and the language assistant can act together as such high-bandwidth partners.
As has been shown in past hypertext interface design studies [11], the primary concern in the creation of an interface for writing A writer may have a predetermined interpretation and future in mind or may write as a means of exploring the interpretative and/or dynamic multiverse. Regardless, a writer must be aware of the multiplicity which deifnes the readers’ and characters’ subjective experiences as the shaper of the meaning and dynamics of the work. The writer thus seeks to simulate and manipulate that multiplicity.
We propose that generative language models in their multiversal modality can serve as an augmentation to and be augmented by the writer’s inherently multiversal imagination. multiverse story is the massive amount of in- written linear and nonlinear stories spanning formation that could be shown to the writer. the equivalent of thousands of pages of often If intuitive user experience is not central to astonishing ingenuity and beauty and surthe design of the program, this information prisingly long-range coherence. Three users will feel overwhelming and functionally pre- have reported a sense of previously unimagvent the user from leveraging the power of- ined creative freedom and expressive power. fered by multiverse access at all. However, it has also become evident that
An efective multiversal interface should much more research and development is necallow the writer, with the aid of a generative essary. In our beta-tests, we’ve found that language model, to expose, explore, and ex- flaws in interface design can easily overploit the interpretational and dynamic multi- whelm or damage a feeling of ownership plicity of a passage. Not only will such a tool over the work produced. Below we will share allow the user to explore the ways in which some of our findings, which represent only a scenario might play out, such an interface the first step in creating a true interface bewill also expose previously unnoticed ambi- tween the creative mind and the machine. guities in the text (and their consequences).
Depending on the design of the interface 4.2. Multiple visualizations and the way the user approaches it, many different human-AI collaborative workflows are possible. Ideally, the interface should give the user a sense of creative superpowers, providing endless inspiration combined with executive control over the narrative, as well as allowing and encouraging the user to intervene to any degree.
We have found that a visual representation of the branching structure of the narrative helps users conceptualize and navigate fractal narratives. This view (called visualize) displays the flow of pasts and futures surrounding each node (Figure 3) and zooming out displays the global structure of the multiverse (Figure 4). The visualize view al4.1. Progress so far lows users to expand and collapse nodes and subtrees, as well as “hoist” any node so that Over the past several months, we have pro- it acts as the root of the tree. Altering the totyped and tested several iterations of mul- topology of the tree, (e.g. reassigning chiltiversal writing tools using GPT-3 as the gen- dren to diferent parents, splitting and mergeration function. ing nodes) is more intuitive for users in the
The demand for a multiversal writing ap- visualize view than the linear view. plication grew from use of GPT-3 as a more In addition to tree-based multiverse visustandard linear writing assistant. It became alization, the read view displays the text of a increasingly clear, as users sought greater in- node and its ancestry in a single-history forteraction bandwidth and more eficient ways mat (Figure 5). to structure and leverage the model’s ideas, that an interface which organizes the model’s 4.3. Multiverse navigation outputs in a branching tree would be more efective. With a generative language model, story
The early results we have seen leave no multiverses can quickly become too large to doubt about the power of language models navigate through node connections alone. To as writing assistants. Our small cohort of assist navigation, we have implemented the ifve beta users have, alongside GPT-3, co- following features: • Indexing by chapters: Chapters are assigned to individual nodes, and all nodes belong to the chapter of the closest ancestor that is the root of a chapter. As a consequence, chapters have the shape of subtrees. • Bookmarks and tags: Bookmarks create a named pointer to a node without enforcing chapter membership. Tags are similar to bookmarks, but can be applied to multiple nodes.
A naive way to automatically generate a multiverse using a language model might involve branching every fixed n tokens. However, this is not the most meaningful way to branch in a story. In some situations, there is essentially one correct answer for what a language model should output next. In such a case, the language model will assign a very high confidence (often >99%) for the the multiverse such as merging interesting top token. Branching at this point would aspects of two separate branches into one. introduce incoherent continuations. Con- The interface should ideally allow the huversely, when the language model distributes man to perform all desired operations with transition probabilities over multiple tokens, maximal ease. Because GPT-3 is so capable branching is more likely to uncover a rich di- of producing high-quality text, some interversity of coherent continuations. face designs make it feasible for the human to
One algorithm to dynamically branch is cultivate coherent and interesting passages to sample distinct tokens until a cumula- through curation alone. We have found that tive probability threshold is met. Adaptive an interface which makes it easy to generate branching allows visualization of the dynam- continuations but relatively dificult to modics of the multiverse: stretches of relative de- ify the content and topology of the resultterminism alternating with divergent junc- ing multiverse encourages a passive worktures (Figure 6). lfow, where the user relies almost exclusively on the language model’s outputs and the 4.5. Reciprocal workflow branching topology determined by the process of generation.
Humans retain an advantage over current While such a passive mode can be fun, language models in our ability to edit writ- resembling an open-ended text adventure ing and perform topological modifications on game, and as well as useful for eficiently ex• Support for nonstandard topologies that are not automatically generated by language models and require human arrangement, such as cycles and multiple parents (§4.7) ploring counterfactuals, the goal of a writ- 4.7. Nonstandard topologies ing interface is to facilitate two-way interaction: the outputs of the language model should augment and inspire the user’s imagination and vice versa.
Thus, we are are developing features to encourage meaningful and unrestrained human contribution such as: The interface supports nodes with multiple parents and allows cyclic graphs (Figure 7). Opportunities to arrange convergent and cyclic topologies, which do not occur if the language model is used passively, encourage human cowriters to play a more active role, for instance, in arranging for sep• Easy ways to edit, move text, and arate branches to converge to a single outchange tree topology come. Multiversal stories naturally invite plots about time travel and weaving timelines, and we have found this feature to unlock many creative possibilities. 4.8. Memory management 4.6. Floating notes • Floating notes to allow saving pas- GPT-3 has a limited context window, which sages and ideas independent from the might seem to imply limited usefulness for tree structure (§4.6) composing longform works like novels, but our users have found that long-range coher• Fine-grained control over language ence is surprisingly easy to maintain. Ofmodel memory (§4.8) ten, the broad unseen past events of the narrative are contained in the interpretational • Interactive writing tools that ofer multiplicity of the present and thus exposed influence over the narrative in ways through generations, and consistent narraother than direct intervention (§4.9) tives are easily achieved through curation. • Program modes which encourage man- In order to reference past information more ual synthesis of content from multi- specifically, often all that is needed is minverse exploration into a single history, imal external suggestion, introduced either for instance by distinguishing between by the author-curator or by a built-in memexploratory and canonical branches ory system. We are developing such a system which automatically saves and indexes story information from which memory can be keyed based on narrative content.
Floating notes are text files which, rather than being associated with a particular node, 4.9. Writing tools are accessible either globally or anywhere in Beyond direct continuations of the body of a subtree. We decided to implement this fea- the story, a language model controlled by enture because users would often have a sepa- gineered prompts can contribute in an openrate text file open in order to copy and paste ended range of modalities. Sudowrite[12] interesting outputs and keep notes with- has pioneered using GPT-3 powered funcout being constrained by the tree structure. tions that, for instance, generate sensory deFloating notes make it easier for the user ex- scriptions of a given object, or prompt for a ert greater agency over the narrative. twist ending given a story summary. [2] B. S. DeWitt, N. Graham, The many
The ability to generate high-quality sum- worlds interpretation of quantum memaries has great utility for memory and as chanics, volume 63, Princeton Univerinput to helper prompts and forms an ex- sity Press, 2015. citing direction for our future research. We [3] D. Deutsch, The Fabric of Reality, Penare exploring summarization pipelines for guin UK, 1998.
GPT-3 that incorporate contextual informa- [4] P. A. Roth, The pasts, History and Thetion and examples of successful summariza- ory 51 (2012) 313–339. tions of similar content. [5] K. Amaral, Hypertext and writing: An overview of the hypertext medium, Retrieved August 16 (1995) 2004. 5. Conclusion [6] E. J. Aarseth, Nonlinearity and literary theory, Hyper/text/theory 52 (1994) The problem of designing good interfaces for 761–780.
AI systems to interact with humans in novel [7] M. Bernstein, Storyspace 1, in: Proceedways will become increasingly important as ings of the thirteenth ACM conference the systems increase in capability. We can on Hypertext and hypermedia, 2002, pp. imagine a bifurcation in humankind’s future: 172–181. one path in which we are left behind once the [8] L. Lagerkvist, M. Ghajargar, Multiverse: machines we create exceed our natural capa- Exploring human machine learning inbilities, and another in which we are uplifted teraction through cybertextual generaalong with them. We hope that this paper can tive literature, in: 10th International further inspire the HCI community to con- Conference on the Internet of Things tribute to this exciting problem of building Companion, 2020, pp. 1–6. the infrastructure for our changing future. [9] M. O. Riedl, R. M. Young, From linear story generation to branching story Acknowledgments graphs, IEEE Computer Graphics and Applications 26 (2006) 23–31. doi:10.
We are grateful to Lav Varshney for his 1109/MCG.2006.56. valuable discussions and helpful feedback [10] M. O. Riedl, V. Bulitko, Interactive narand to Michael Ivanitskiy and John Balis for rative: An intelligent systems approach, their feedback and help compiling this arti- Ai Magazine 34 (2013) 67–67. cle. In addition we would like to thank Miles [11] S. Jordan, ‘an infinitude of possible Brundage and OpenAI for providing access worlds’: Towards a research method to GPT-3. for hypertext fiction, New Writing 11 (2014) 324–334. [12] A. Gupta, J. Yu, References https://www.sudowrite.com/, 2021. URL: https://www.sudowrite.com/.