News from the ACM Publication Board
AI Initiatives: The ACM is trying to expand its presence in AI in order to better meet the needs of the AI community. A challenge is that while we publish a lot of research papers involving AI, the highest-profile AI-specific venues are not ACM venues. However, there are also opportunities: ACM has a presence in virtually all areas of computer science (that’s you–the SIGs!), and as AI becomes intertwined with research and practice across computer science, an organization with breadth potentially has a lot to offer in supporting interdisciplinary AI work. ACM has recently started 4 journals related to AI:
* Trans. Probabilistic ML
* Trans. AI for Science
* Trans. AI Security and Privacy
* AI Letters
Additional publication venues related to AI are being considered. In general, input on how Publications can better support AI in ACM is very welcome, I will pass it on.
AI Summit: ACM is planning an AI Summit meeting to be held roughly a year from now. Location is TBD but probably in the US. It will be open to attendees from the public but will consist mostly of invited talks and panels. Topics are also TBD but may include AI research, AI in society, and AI’s relation to other areas of CS. The goals I heard are threefold:
* For ACM: Showcase how ACM can contribute to AI
* For attendees: Learn about interdisciplinary AI
* For both: Help ACM and the research community form a strategy around AI
AI in ACM Publications Processes and Tools: We are looking at uses of AI in ACM’s publications and reviewing systems to help identify policy violations–will likely have more to share on this in the not-too-distant future. Also, AI features such as short summaries of articles (different from abstracts in that they are aimed at a more general audience), and AI-generated podcasts that summarize a session of a conference are being prototyped for incorporation in a “Premium” version of the ACM Digital Library, available to people in ACM OPEN institutions.
LLM training on ACM papers: this is a continuing area of discussion. The ACM does not currently license content from the digital library to train LLMs. However, there is potential benefit to having knowledge from ACM papers included in results to LLM queries, and if those results include citations, then they could increase the impact of and citations to papers by ACM authors. ACM Publications is committed to having careful and meaningful dialogue with the SIGs, our authors, and the community before making any changes to the status quo in this area.
Publications Strategy: The Board discussed the following questions:
* Is the current model of journal and conference publications serving the community well, or is there a better model? For example, an alternative model (suggested by Yannis Ioannidis, ACM President) is to disaggregate: people would publish in the ACM DL as people do today in arXiv, then apply to be reviewed for acceptance into a journal that is an “overlay” over the DL, and separately apply to present the work at a conference.
* Should there be an ACM venue in every area of CS and AI, whether or not there are competing titles from other publishers?
* Strategically, should we continue to launch and maintain high-quality niche journals? Or should we focus on broad-based journals across entire domains (imagine: “Transactions on Machine Learning”)?
* Is the journal the venue of the future or should we focus elsewhere? Should we accelerate the transition to a journal-first model for conferences?
Thoughts/feedback on these from the SIGs is welcome.
ACM and IEEE: I know a lot of people are worried about joint ACM/IEEE conferences under the new ACM OPEN model, as IEEE has been unwilling to go full Open Access along with ACM. However, a compromise is in the works. This is not ready to publicize, but the likely outcome is that in years where ACM is the publisher of record, papers will be expected to have either ACM OPEN coverage or pay an APC or get a waiver, and then would be freely available in the ACM Digital Library. In years where IEEE is the publisher, papers would be freely available if authors pay an APC to IEEE; other papers would only be available in the Premium (ACM OPEN subscribing institution) version of the ACM Digital Library.