“The less I type, the better”: How AI Language Models can Enhance or Impede Communication for AAC Users
Open accessThe DOI matches its Crossref record; 38 of the 38 references it cites were found in a registry or an index; 5 of the 24 sources its claims cite were checked for support.
Abstract
Users of augmentative and alternative communication (AAC) devices sometimes find it difficult to communicate in real time with others due to the time it takes to compose messages. AI technologies such as large language models (LLMs) provide an opportunity to support AAC users by improving the quality and variety of text suggestions. However, these technologies may fundamentally change how users interact with AAC devices as users transition from typing their own phrases to prompting and selecting AI-generated phrases. We conducted a study in which 12 AAC users tested live suggestions from a language model across three usage scenarios: extending short replies, answering biographical questions, and requesting assistance. Our study participants believed that AI-generated phrases could save time, physical and cognitive effort when communicating, but felt it was important that these phrases reflect their own communication style and preferences. This work identifies opportunities and challenges for future AI-enhanced AAC devices.
Overview
CoChat Summary
TL;DR
This CHI 2023 paper studies how 12 adult AAC users respond to AI language model–generated “speech macros” that expand short inputs into full phrases for different communication tasks. Participants saw clear potential to reduce typing, physical effort, and cognitive effort, but only if the output matched their tone, identity, and privacy needs.
Key findings
- Across the study, participants found the macros generally useful, with “Turn words into Requests” rated most often as extremely or very useful, followed by “Reply with Background Information” and then “Extend Reply.”
- Participants valued the potential to save time and effort; one participant said the extend-reply macro could help them “socialize and network more,” and another said it could reduce cognitive effort in figuring out what to say.
- The model sometimes produced plausible but inappropriate or inaccurate output: for example, with input “transfer” it suggested money-transfer phrases, and with “suction” it suggested unrelated content like “a suction cup that needs to go on the wall.”
- Participants wanted output to reflect their personal style and identity; they often described suggestions as too bland, impersonal, abrupt, or robotic, and some worried automated phrases could affect relationships or make them seem less authentic.
Methods
- The authors ran a remote user study with 12 adult expert AAC users in the United States, the United Kingdom, or Canada; participants used a variety of AAC access methods including eye gaze, switch scanning, direct selection, touch, mouse/joystick, keyboard, and toes.
- Participants tested three speech macro prototypes built with LaMDA: Extend Reply, Reply with Background Information, and Turn Words into Requests.
- Data came from a 90-minute video call plus a post-study survey; recordings were transcribed, combined with chat logs and notes, and analyzed qualitatively using thematic analysis and affinity diagramming by two researchers.
Limitations
- The paper states that participants were limited to 5–10 inputs during the session, so feedback was based on a short, constrained exposure rather than extended use.
- The prototype used a generic dialog-based language model and did not customize speech macros with user-specific data; the paper says this means it does not provide a clear picture of how well current language models could perform for AAC suggestions.
- The study included participants with typical language skills and no aphasia, so this is an evident limitation in scope: the findings may not transfer to AAC users with language comprehension or production impairments.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures



Verification
Claims the paper makes
5 Valid · 3 Not valid · 12 Not checked · 83 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Preliminary research with simulated user data has shown that LLMs can retrieve contextually relevant sentences [31]”
What the sources say
- Partly supportedcites [31] [31] Junxiao Shen, Boyin Yang, John J Dudley, and Per Ola Kristensson. 2022. KWickChat: A Multi-Turn Dialogue System for AAC Using Context-Aware Sentence Generation by Bag-of-Keywords. In 27th International Conference on Intelligent User Interfaces (Helsinki, Finland) (IUI '22) . Association for Computing Machinery, New York, NY, USA, 853-867. https://doi.org/10.1145/3490099.3511145 - Ed Chi, and Quoc Le. 2022. LaMDA: Language Models for Dialog Applications. https://doi.org/10.48550/ARXIV.2201.08239 - sourceExperimental — accuracy not yet measured
Sections 7 and 8 evaluate a language model using a surrogate user and simulated conversational context, and report that context helps it produce better responses. But Section 3 distinguishes the paper’s GPT-2 sentence-generation model from sentence retrieval: the LLM generates sentences rather than retrieving them. Page numbers are not available in our copy. (LLM, conf=0.86)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Current LLMs are able to generate text that is indistinguishable from text written by a human [8]”
What the sources say
- Partly supportedcites [8] [8] Elizabeth Clark, Tal August, Sofa Serrano, Nikita Haduong, Suchin Gururangan, and Noah A Smith. 2021. All That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) . Association for Computational Linguistics, Online, 7282-7296. - sourceExperimental — accuracy not yet measured
The Results (§2.4; page unavailable) report that untrained evaluators distinguished GPT-3 text from human-authored text at chance level across stories, news, and recipes. The claim generalizes this finding to current LLMs broadly: the same section reports that evaluators distinguished GPT-2 text from human-authored text significantly better than chance. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our study participants believed that AI-generated phrases could save time, physical and cognitive efort when communicating”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Many phrases were either considered as very appropriate or were phrases that were close to something the participant would say but that needed a little more editing to get it right.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We found the system was not able to suggest relevant requests related to access or health needs.”
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
35 more claims
Create a free account to see all 103 claims and to verify the 1 reference still open.
- Probably not a substantial claim. 47
- It may state two separate claims as one. 31
- The quote only partly states this claim. 2
Citations
Cites
Using thematic analysis in psychology
Braun, V., & Clarke, V. (2006) · Qualitative Research in Psychology · doi:10.1191/1478088706qp063oa · Cited by 194,414
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325 · cited 1×
“Recently, advances in large, neural language models (LLMs) such as GPT-3 [6] and BERT [10] have created new opportunities for improving the usability and efciency of AAC devices.”
BackgroundQualitative Data Analysis: An Expanded Sourcebook
(1994) · Journal of Environmental Psychology · doi:10.1016/s0272-4944(05)80231-2 · Cited by 48,420
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Computer-Mediated Communication: Impersonal, Interpersonal, and Hyperpersonal Interaction
WALTHER, J. B. (1996) · Communication Research · doi:10.1177/009365096023001001 · Cited by 4,798
LaMDA: Language Models for Dialog Applications
Thoppilan, R., De Freitas, D., Hall, J., Shazeer, N., Kulshreshtha, A., Cheng, H.-T., Jin, A., Bos, T., Baker, L., Du, Y., et al. (2022) · arXiv · doi:10.48550/arxiv.2201.08239 · Cited by 778 · cited 1×
“Informed by prior work that uncovered challenges in AAC-based social interactions [16, 20, 35], we created multiple prompts using the transformer-based large language model LaMDA developed for dialogue applications [32].”
MethodRe-examining Whether, Why, and How Human-AI Interaction Is Uniquely Difficult to Design
Yang, Q., Steinfeld, A., Rosé, C., & Zimmerman, J. (2020) · Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3313831.3376301 · Cited by 622
Lee, C. P. (2007) · Computer Supported Cooperative Work (CSCW) · doi:10.1007/s10606-007-9044-5 · Cited by 413 · cited 1×
“…to communicating technology capabilities and limitations to users, designers, and technologists has been the use of boundary objects [40], representations of abstract ideas in the form of interactive prototypes or other artifacts that facilitate communication among diferent stakeholders [23].”
BackgroundOh, C., Song, J., Choi, J., Kim, S., Lee, S., & Suh, B. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3174223 · Cited by 301 · cited 1×
“Other approaches have developed open-ended working prototypes to explore what tasks an AI system can and cannot do for specifc users [26, 28].”
BackgroundAll That’s ‘Human’ Is Not Gold: Evaluating Human Evaluation of Generated Text
Clark, E., August, T., Serrano, S., Haduong, N., Gururangan, S., & Smith, N. A. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.565 · Cited by 290 · cited 1×
“Current LLMs are able to generate text that is indistinguishable from text written by a human [8], potentially enabling AAC users to generate human-level speech with minimal efort.”
Background
Research works printed but not linked: 4
- Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei. 2020. Language Models are Few-Shot Learners. In Advances in Neural Information Processing Systems, H. Larochelle, M. Ranzato, R. Hadsell, M.F. Balcan, and H. Lin (Eds.). Vol. 33. Curran Associates, Inc., 1877–1901. https://proceedings.neurips.cc/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf
- 10.18653/v1 · doi:10.18653/v1
- 10.18653/v1 · doi:10.18653/v1
- 10.5555/2392855.2392859 · doi:10.5555/2392855.2392859
Cited by
Yildirim, N., Richardson, H., Wetscherek, M. T., Bajwa, J., Jacob, J., Pinnock, M. A., Harris, S., Coelho De Castro, D., Bannur, S., Hyland, S., et al. (2024) · Proceedings of the CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3642013 · Cited by 82
“Recent literature highlights prompt prototyping as a potential research direction for experience prototyping with LLMs and generative AI capabilities [63, 103, 131].”
"I look at it as the king of knowledge": How Blind People Use and Understand Generative AI Tools
Adnin, R., & Das, M. (2024) · The 26th International ACM SIGACCESS Conference on Computers and Accessibility · doi:10.1145/3663548.3675631 · Cited by 73
“…in their regular work practices, extending prior research that investigated disability representation and biases in GenAI [ 25, 29, 60] and potentials of GenAI to support access needs of people with neurodivergence [32, 46, 86] and other disabilities like chronic illness and aphantasia [30].”
BackgroundJang, J., Moharana, S., Carrington, P., & Begel, A. (2024) · Proceedings of the CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3642894 · Cited by 64
“Others have suggested its potential for supporting communication for people with disabilities, such as enhancing experiences for AAC users [87], writing AI-assisted emails for adults with dyslexia [28] and even speech-language pathologists in delivering therapy to autistic individuals in clinical…”
BackgroundTang, Y., Chen, L., Chen, Z., Chen, W., Cai, Y., Du, Y., Yang, F., & Sun, L. (2024) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3642899 · Cited by 101
“Valencia et al. [14] utilized the ability of LLMs to extract, expand, and generate information to support the specific communication needs of AAC users.”
BackgroundInfluentialPang, R. Y., Schroeder, H., Smith, K. S., Barocas, S., Xiao, Z., Tseng, E., & Bragg, D. (2025) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713726 · Cited by 146
“We found diverse accessibility contexts, including the blind or low-vision (BLV) community [151, 180], people with autism [70], learning disabilities involving Augmentative and Alternative Communication (AAC) [156], andsituationalimpairments[106],aswellaspapers onolderadults[172].”
Metadata
- Authors
- Valencia, S., Cave, R., Kallarackal, K., Seaver, K., Terry, M., & Kane, S. K.
- Affiliations
- Google (United States) (US)
- University College London (GB)
- MGH Institute of Health Professions (US)
- Venue
- Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems
- Publisher
- ACM
- Published
- 2023-04-19
- Pages
- 1-14
- OpenAlex
- W4366551359
- Cited by
- 97
- References
- 29
- Citations by year
- 2026: 25 · 2025: 43 · 2024: 26 · 2023: 3
- FWCI
- 66.89
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- CHI Conference on Human Factors in Computing Systems (publishedVersion, cc-by)
- UCL Discovery (University College London) (submittedVersion, cc-by)
- Topics
- Health Sciences › Health Professions › Occupational Therapy › Assistive Technology in Communication and Mobility
- Physical Sciences › Computer Science › Human-Computer Interaction › Gaze Tracking and Assistive Technology
- Social Sciences › Human Factors and Ergonomics › Digital Accessibility for Disabilities
- Keywords
- augmentative and alternative communication, large language models, communication effort, user preferences, communication styles, real-time communication, assistive technology
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-01, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-01 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: “The less I type, the better”: How AI Language Models can Enhance or Impede Communication for AAC Users. https://app.cochat.ai/papers/10.1145/3544548.3581560
- The paper
- Valencia, S., Cave, R., Kallarackal, K., Seaver, K., Terry, M., & Kane, S. K. (2023). “The less I type, the better”: How AI Language Models can Enhance or Impede Communication for AAC Users. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3544548.3581560
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.