Elamur Research⌕
☰
home / papers / 2026 / Launch and Aftermath: Contrasting Social Media Responses to Chatbot Releases. The Cases of Meta's Galactica and OpenAI's ChatGPT

Launch and Aftermath: Contrasting Social Media Responses to Chatbot Releases. The Cases of Meta's Galactica and OpenAI's ChatGPT

Paper · 2026 · Maximilian Weber and Johannes B. Gruber

Summary

In November 2022, Meta's Galactica and OpenAI's ChatGPT launched about two weeks apart. Galactica was pulled within days, while ChatGPT became a mass-market success. The authors treat this as a natural comparison and ask three questions. How did Twitter conversations about the two launches differ? Did talk of harms grow over time for Galactica? Did positioning for science rather than general use change how the models were received?

They gathered English tweets through the Twitter Research API and filtered out unrelated "Galactica" mentions. They then combined three methods:

  • a Twitter-tuned RoBERTa sentiment classifier;
  • zero-shot harm/risk labelling with Llama 3.3;
  • an LLM-driven topic pipeline with human-in-the-loop refinement.

Negative sentiment toward Galactica rose sharply within days, and harm-focused tweets came to make up most of its discussion. ChatGPT's negativity stayed fairly flat, and its discussion mixed hype, experimentation and criticism. The authors argue the gap reflects how each model was framed, not technical differences. Presenting Galactica as a trustworthy scientific tool made its hallucinations look far more serious. The lesson concerns expectation management in AI releases.

Key points

  • Data: 2,077 Galactica tweets (Nov 15–Dec 15, 2022) and 107,726 ChatGPT tweets (Nov 30–Dec 7, 2022), with 283 mentioning both. The main analyses cover each model's first eight days.
  • Relevance filter: Off-topic tweets (e.g., Battlestar Galactica) were removed using snowflake-arctic-embed2 embeddings (via rollama) and a LASSO logistic regression. The classifier was trained on 250 hand-labelled tweets and reached accuracy 0.908 and macro F1 0.873.
  • Sentiment: Measured with cardiffnlp/twitter-roberta-base-sentiment-latest.
    • Galactica's negative share went from 5.1% on day 1 to 35.6% on day 3 and peaked at 49.1% on day 5.
    • ChatGPT started at 12.0% and stayed around 17–21%.
    • Differences were significant (Fisher's exact test, BH-adjusted) on every day except launch day.
  • Harm/risk annotation: Zero-shot labelling with Llama-3.3-70B-Instruct (4-bit NF4).
    • Agreement with two human annotators: κ 0.688 and 0.766, macro F1 0.843 and 0.883. The humans agreed with each other at κ = 0.67.
    • Harm-related Galactica tweets rose from 8.3% on day 1 to 57% by day 5.
  • Topics: Assigned with Llama 3.3 70B via Ollama, following an adapted Pham et al. (2024) pipeline. ChatGPT was analysed on a 1,640-tweet subsample.
    • ChatGPT discussion was led by Innovation, Technical Details, Hype/Excitement and Criticism.
    • Galactica discussion was led by Criticism, Scientific Utility and AI Safety/Risks.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Johannes Gutenberg University MainzMainz · Germany6 works · 665 citations🔬GESIS - Leibniz Institute for the Social SciencesMannheim · Germany10 works · 5 citations

 Fields, topics and concepts

📚 Shelves 7The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksSocial Sciences · 1,772 worksSafety Research · 543 worksGeneral Social Sciences · 227 works

🧭 Topics 1Research subjects and languages the work is about.Dialogue systems · 624 works

Cite

Cite as Maximilian Weber and Johannes B. Gruber. 2026. Launch and Aftermath: Contrasting Social Media Responses to Chatbot Releases. The Cases of Meta's Galactica and OpenAI's ChatGPT. Proceedings of the Seventh Workshop on Natural Language Processing and Computational Social Science.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.nlpcss-1.5

@inproceedings{weber2026launch,
  title = {{Launch and Aftermath: Contrasting Social Media Responses to Chatbot Releases. The Cases of Meta's Galactica and OpenAI's ChatGPT}},
  author = {Maximilian Weber and Johannes B. Gruber},
  booktitle = {Proceedings of the Seventh Workshop on Natural Language Processing and Computational Social Science},
  year = {2026},
  doi = {10.18653/v1/2026.nlpcss-1.5},
  url = {https://doi.org/10.18653/v1/2026.nlpcss-1.5}
}

References

14 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared 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 M. Ziegler, Jeffrey Wu, Clemens Winter,… find
  2. Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Christiano, Jan Leike, and Ryan Lowe. 2022. Train…
  3. Chau Minh Pham, Alexander Hoyle, Simeng Sun, Philip Resnik, and Mohit Iyyer. 2024. TopicGPT: A prompt-based topic modeling framework. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Pape…
  4. Nicolas Chartier-Edwards, Etienne Grenier, and Valentin Goujon. 2024. Galactica’s dis-assemblage: Meta’s beta and the omega of post-human science. AI & SOCIETY. find
  5. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, and 542 o… — The Llama 3 Herd of Models
  6. Johannes B. Gruber and Maximilian Weber. 2024. rollama: An R package for using generative large language models through Ollama. arXiv preprint. ArXiv:2404.07654 [cs]. — [arXiv:2404.07654](https://arxiv.org/abs/2404.07654) find
  7. Ratanond Koonchanok, Yanling Pan, and Hyeju Jang. 2024. Public attitudes toward chatgpt on twitter: sentiments, topics, and occupations. Social Network Analysis and Mining, 14(1):106. find
  8. Jianghan Liu, Ziyu Shang, Wenjun Ke, Peng Wang, Zhizhao Luo, Jiajun Liu, Guozheng Li, and Yining Daniel Loureiro, Francesco Barbieri, Leonardo Neves, Luis Espinosa Anke, and Jose Camacho-Collados. 2022. TimeLMs: Diachronic language models from Twitter. In Proceedings of the 60th Annual Meeting of th…
  9. Adrian Rauchfleisch, Joshua Philip Suarez, Nikka Marie Sales, and Andreas Jungherr. 2025. Winning and losing with Artificial Intelligence: What public discourse about ChatGPT tells us about how societies make sense of technological change. Telematics and Informatics, 103:102344. find
  10. Yash Sharma. 2025. MALTopic: Multi-Agent LLM Topic Modeling Framework. In 2025 IEEE World AI IoT Congress (AIIoT), pages 0707–0712, Seattle, WA, USA. IEEE. find
  11. Ross Taylor, Marcin Kardas, Guillem Cucurull, Thomas Scialom, Anthony Hartshorn, Elvis Saravia, Andrew Poulton, Viktor Kerkez, and Robert Stojnic. 2022. Galactica: A large language model for science. Preprint, arXiv:2211.09085.
  12. Cascha van Wanrooij and Omendra Kumar Manhar. 2024. Topic Modeling for Small Data using Generative LLMs. In Proceedings of the 36th Benelux Conference on Artificial Intelligence (BNAIC) and the 33rd Belgian-Dutch Conference on Machine Learning (BeNeLearn) 2024, Utrecht, Netherlands. find
  13. Maximilian Weber. 2024. Social Group Differences in the Social Media Discussion about ChatGPT and Bing Chat. In ACM Web Science Conference, pages 114–118, Stuttgart Germany. ACM. find
  14. Yann LeCun [@ylecun]. 2023. Galactica, the LLM for scientists from Meta, was released a couple of weeks before ChatGPT but was taken down after... [Post]. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/weber-2026-launch-aftermath-contrasting-social-media/