# Next Steps for Human-Centered Generative AI: A Technical Perspective

**Authors:** Xiang 'Anthony' Chen, Jeff Burke, Ruofei Du, Matthew K. Hong, Jennifer Jacobs, Philippe Laban, Dingzeyu Li, Nanyun Peng, Karl D. D. Willis, Chien-Sheng Wu, Bolei Zhou  
**arXiv:** [2306.15774](https://arxiv.org/abs/2306.15774)  
**Date:** 2023-06-27  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2306.15774v1/

## Key points

- Authors present a roadmap for future technical research on Human-centered Generative AI (HGAI)
- Roadmap aims to define and propose next steps for HGAI from a technical perspective
- Three iterations of discussions involving researchers from academia and industry in various fields
- First iteration involved individual brainstorming discussions to generate ideas for HGAI research agenda
- Notes from each discussion were summarized into a list of research agenda topics with brief descriptions
- Topics were aggregated to form a comprehensive roadmap for future technical research on HGAI
- Main contribution of the paper is the roadmap itself, spanning three levels: aligning with human values, accommodating humans' expression of intents, and augmenting humans' abilities in a collaborative workflow
- Roadmap aims to draw interdisciplinary research teams and maintain a coherent big picture of the future work landscape
- Paper provides valuable insights into next steps for technical research on Human-centered Generative AI by leveraging expertise across different disciplines and fostering interdisciplinary collaboration.

## Summary

In this paper, the authors present a roadmap for future technical research on Human-centered Generative AI (HGAI). They aim to define and propose next steps for HGAI from a technical perspective through iterative, cross-disciplinary discussions. The authors conducted three iterations of discussions involving researchers from academia and industry in fields such as technical HCI research, machine learning, natural language processing, computer vision, and computer graphics. The first iteration involved individual brainstorming discussions between the first author and each participant. The purpose was to generate as many ideas as possible to establish the breadth of an HGAI research agenda. Ideas were generated on defining HGAI, identifying HGAI-related problems, and proposing new topics for future research. After the individual brainstorming discussions, the authors summarized the notes from each discussion into a list of research agenda topics with brief descriptions. These topics were then aggregated to form a comprehensive roadmap for future technical research on HGAI. The main contribution of this paper is the roadmap itself, which spans three levels: aligning with human values, accommodating humans' expression of intents, and augmenting humans' abilities in a collaborative workflow. The roadmap aims to draw interdisciplinary research teams to a comprehensive list of emergent ideas in HGAI while maintaining a coherent big picture of the future work landscape. Overall, this paper provides valuable insights into the next steps for technical research on Human-centered Generative AI by leveraging expertise across different disciplines and fostering interdisciplinary collaboration.