/ai

An "AI submissions" journal is reportedly in the works · Digg

An "AI submissions" journal is reportedly in the works

The user describes the journal as a project that will "parallel arXiv."

TLDR

In a reply, a user says an "AI submissions" journal is being made and will "parallel arXiv."

Combined views

12.7K

16 Sources, first seen 4d ago

85 likes

An "AI submissions" journal is reportedly in the works

The user describes the journal as a project that will "parallel arXiv."

Combined views

12.7K

16 Sources, first seen 4d ago

85 likes

4d ago

first seen 4d ago

10 comments

11 saves

8 reposts

4d ago

first seen 4d ago

TLDR

In a reply, a user says an "AI submissions" journal is being made and will "parallel arXiv."

10 comments

11 saves

8 reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

16 Sources

@yaroslavvb@tdietterich There is an "AI submissions" journal being made that will parallel arXiv. Cc @bohannon_bot @Kangwook_Leesimilarly, the notion of "human-assisted" won't exist in a few years. it's either 100% human-done (hand-made papers) or 100% ai-done.@ChenhaoTanA separate repository makes a lot of sense for the AI-written/AI-verified. I think emphasizing human-written in the first scenario is distracting. We should reduce the incentive to seek publication so that the first category is for people who think publishing would still make sense.@DimitrisPapailRT @Kangwook_Lee: similarly, the notion of "human-assisted" won't exist in a few years. it's either 100% human-done (hand-made papers) or…@adamwslave@krismicinski It’s really important that you learn and practice the skills regarding synthesis of research. Having an LLM do the research assistant bit is fine, but if you’re unable to synthesize/interpret what’s gathered, why are you in academia in the first place?@zehavocRT @adamwslave: @krismicinski It’s really important that you learn and practice the skills regarding synthesis of research. Having an LLM d…@anmarasovic@TuhinChakr I'm worried who checks review quality – even our meta-reviewers aren't reliable... I still think we need desk reject of anything that smells like AI slop. I'd like that an expert reviewer checks the intro of any Pangram-flagged paper and reject if they can't understand it.@TuhinChakr@anmarasovic Or we develop a model that can identify scientific slop. This could be valuable if combined with AI detection @anmarasovic@TuhinChakr I'd even introduce loosing tokens for those who submit AI slop author responses. After CoLM, I won't put myself through any more reading of AI slop than I minimally need to, so my service contribution is much lower than ever before in my 10 years giving service to conferences.@aran_nayebiarXiv should remain a repository of timestamped results, and not serve the job of an editor/gatekeeper. This is still an incredibly valuable service imho b/c at this rate, only an LLM reviewer can keep up with the scale of submissions.@delipraoalso, if students produce increasingly larger volume of output due to agents, the due diligence workload on the PI becomes hard to handle and error prone. one solution is to pair junior students with senior students, reinventing middle managers, but that also means you have a steady supply of students and grant money to have a pipeline of students at different experience levels.@tallinzenInterestingly, AI slop papers have made it easier on average to review for ML conferences because many papers quickly turn out to not merit a careful review, so you only have to actually review 25-50% of your batch.@TuhinChakrRT @JessicaHullman: Sad but not surprising. It’s pretty obvious that at this point in time, heavily AI-generated papers are less likely t…@1a1a11aUnpopular opinion—we should not publish system design papers any more (given that most papers have "untold story", instead, we should focus on (1) the problem (2) why the solution from AI does not work. How many disagree?@Dr_AtoosaTraditional AI detection tools (like Pangram) merely police token-level generated products (e.g., papers). We need systems that actively reward authorship accountability, critical thinking, rigor, and clarity in order to sustain scientific research. 2/11 @labenzPrakash on AI:AM: the traditional PhD — unassisted by AI, adding something new to humanity's knowledge — is 'kind of over,' and it happened in about a month. The pioneer role goes to the AI. What's left for the human: understand, verify, correct. https://x.com/i/broadcasts/1qKVmyRRNQmxB?t=486

Today's Rank

—

Not ranked yet

Today's Rank

—

Not ranked yet