Guide
The world's top AI conferences
What universities and AI labs look for
College admissions
“A full research paper or publication is strongly preferred.”
“Full research papers are generally the most helpful.”
“About 42% of our admitted students from the recent class submitted research papers and publications as part of their supplemental materials.”
Graduate school
“Meaningful research experience is often the most important factor in admission decisions.”
“Our Admissions Committee wants to know about your ability to do independent research.”
“A weakness in one area (say, GPA) can be compensated by demonstrated strength in another (say, real implementation experience, or research publications, or recommendation letters).”
AI labs and tech companies
“Track record of publications at NeurIPS, ICLR, ICML.”
“Publication record in relevant conferences (e.g., NeurIPS, ICML, ICLR, AAAI, CVPR, ICCV, ECCV, ACL, EMNLP, etc.)”
“Publications at peer-reviewed venues (NeurIPS, ICML, ICLR, ACL, EMNLP, or similar)…”
How AI venues rank
Google Scholar ranks journals and conferences by how often their recent papers are cited. It keeps a separate list for each field. The ranked examples below come from its Artificial Intelligence, Computational Linguistics and Computer Vision lists.
Google Scholar orders each list by a venue's h5-index. It is the largest number h for which h of the venue's papers from the last five complete years (2020 to 2024) have each been cited at least h times. NeurIPS has an h5-index of 371. That means at least 371 NeurIPS papers from those years have each been cited 371 times or more. A higher number means more of the venue's recent work is widely cited by other researchers.
The Tier 1 conferences are the top three in each of Google Scholar's artificial intelligence, language and computer vision lists: NeurIPS, ICLR and ICML; ACL, EMNLP and NAACL; CVPR, ECCV and ICCV.
- 1
Tier 1 conferences
The top three in each Google Scholar list. Full-length papers, accepted through double-blind peer review.
- #1CVPR
- #2ECCV
- #3ICCV
Robotics has its own leading conferences: ICRA and CoRL. Algoverse teams do not publish there, because most robotics research needs a physical lab.
- 2
Other established AI conferences
Ranked just below Tier 1. Full-length papers, accepted through double-blind peer review.
Venue Google Scholar rank AAAI #4 in Artificial Intelligence WACV #6 in Computer Vision & Pattern Recognition COLING #5 in Computational Linguistics EACL #6 in Computational Linguistics COLM The Conference on Language Modeling. First met in 2024, so not yet in Google Scholar’s five-year rankings. - 3
Venues that carry little weight in AI research
Reviewed by students, by the program itself or by people outside AI.
Student journals and conferences, symposia and journals a program runs for its own students, and journals that charge authors to publish. None of them appears in Google Scholar’s Artificial Intelligence, Computational Linguistics or Computer Vision lists. The ones Scholar Metrics tracks have an h5-index between 3 and 28, against 371 for NeurIPS.
Student journal
Journal of Emerging Investigators
Publishes middle and high school authors aged 13 or older, each with an adult senior author. Charges $49 at submission.
Source: JEI, submission FAQ. Not listed in Google Scholar Metrics.
Student journal
Curieux Academic Journal
Describes itself as a youth-led nonprofit. Publishes high school and middle school students.
Source: curieuxacademicjournal.com. Not listed in Google Scholar Metrics.
Student journal
National High School Journal of Science
Publishes research by high school students. Its reviewers are graduate students, postdocs and scientists.
Student journal
International Journal of High School Research
Publishes high school authors. Reviewed by faculty and professionals, by its own description.
Source: ijhsr.terrajournals.org. h5-index 3 in Google Scholar Metrics.
Regional conference
International Conference on ICT Convergence
Run by the Korean Institute of Communications and Information Sciences on Jeju Island. IEEE is a technical co-sponsor. Not in Google Scholar’s AI, language or vision lists.
Program symposium
A program’s own symposium or journal
The presenters are the program’s own students and the program selects them. Nothing outside the program lists the paper.
Competition
Science fairs
Regeneron ISEF calls itself the world’s largest pre-college science competition. Judges award prizes. No paper is published or reviewed by the field.
Source: Society for Science.
- 4
Preprints and self-publishing
No review.
arXiv and similar servers post a paper without review. So does a PDF on a program’s site, a personal page or a program’s own preprint server. A preprint shows that the paper exists. The conference listing shows whether it was accepted anywhere.
Preprint server
arXiv
“The arXiv moderation process is not a peer-review process,” in arXiv’s words. Moderators check that a submission is on topic and in scholarly form.
Source: arXiv, moderation.
Self-published
A PDF on a website
A paper posted on a program’s site, a personal page or a document-sharing service has not been reviewed.
How to check any venue
Is it in Google Scholar’s rankings?
Search the venue’s name in Google Scholar Metrics. NeurIPS has an h5-index of 371. A venue with no entry has too few cited papers to be listed at all. Google Scholar Metrics.
Who reviews?
Look for the program committee or reviewer list. At an AI conference the organizers assign researchers in the topic. At a student journal the reviewers are often students. Some journals do not say who reviews.
What share of submissions is accepted?
Conferences publish the figure in their opening remarks or on OpenReview. The main AI conferences accept between a quarter and a third. If a venue publishes no figure, ask the program for one.
Does the author pay to publish?
A submission or publication fee is on the journal’s guidelines page. AI conferences charge nothing to submit. Presenting requires a conference registration.
Would an AI researcher cite it?
Search the venue’s name on Google Scholar and read who cites its papers. Papers at the first two levels above are cited by other researchers in the field.
Ranks and h5-index: Google Scholar Metrics, Artificial Intelligence, Computational Linguistics and Computer Vision & Pattern Recognition lists, retrieved September 28, 2026. Venues in levels 3 and 4 were looked up in Google Scholar Metrics on September 28, 2026. Each entry links to the lookup.
In computer science the conference is the main venue. A Computing Research Association guide to evaluating faculty for tenure says that “the premier conferences are generally more selective than the premier journals.” Read the guide.
Where modern AI was first published
The papers behind today's AI products were first published at these conferences. Each title below links to the paper on the conference's own site.
1987 · NIPS
The first NIPS meeting
The conference now called NeurIPS met for the first time, with 90 papers. It has met every year since and is now the most-cited venue in AI.
2012 · NeurIPS
ImageNet Classification with Deep Convolutional Neural Networks
Began the deep learning era. It showed that neural networks could recognize images far better than any earlier method, and modern photo search and medical imaging AI descend from it.
2017 · NeurIPS
Attention Is All You Need
Introduced the Transformer, the design that ChatGPT, Claude and Gemini are built on.
2019 · NAACL
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
The language model Google added to Search that October. It won Best Long Paper at NAACL 2019.
Google applied it to one in ten English searches in the US. Google, October 25, 2019.
2020 · NeurIPS
Language Models are Few-Shot Learners
The GPT-3 paper. It showed that one model could take on new tasks from a few examples in a prompt, and it won a best paper award at the conference.
2022 · NeurIPS
Training Language Models to Follow Instructions with Human Feedback
Introduced the training method OpenAI then used to build ChatGPT: people rank a model’s answers, and the model learns from those rankings.
Main conference and workshops
Each Tier 1 conference publishes full-length, peer-reviewed papers in two places: the main conference and the workshops it hosts.
Main conference
The conference's broadest program, covering every area of the field. NeurIPS, ICML and ICLR accept between a quarter and a third of submissions.
Workshops
Focused programs on one research area, organized and reviewed by leading researchers in that area and presented at the conference. word2vec first appeared at an ICLR 2013 workshop, knowledge distillation at a NIPS 2014 workshop and DeepMind's first Atari paper at a NIPS 2013 workshop.
Look up any paper in 30 seconds
- Ask for the exact title of the paper.
- Search the title on OpenReview (openreview.net) or the ACL Anthology (aclanthology.org). These are the public systems the conferences themselves use.
- The listing names the conference, the track and every author. If a paper cannot be found there or on arXiv, ask why.
This works for any program's papers. Try it on one of our students' papers: Scratchpad Thinking: Alternation Between Storage and Computation in Latent Reasoning Models.
Names and terms
Conferences
- NeurIPS
- Conference on Neural Information Processing SystemsThe largest machine learning conference. Held every December since 1987. First in Google Scholar’s Artificial Intelligence list.
- ICML
- International Conference on Machine LearningHeld every July. One of the three general machine learning conferences along with NeurIPS and ICLR. Third in Google Scholar’s Artificial Intelligence list.
- ICLR
- International Conference on Learning RepresentationsHeld every spring. Anyone can read its reviews on OpenReview. Second in Google Scholar’s Artificial Intelligence list.
- ACL
- Annual Meeting of the Association for Computational LinguisticsThe main conference for research on language and AI. Chatbots come out of this area. First in Google Scholar’s Computational Linguistics list.
- EMNLP
- Conference on Empirical Methods in Natural Language ProcessingRun by the same association as ACL and held each fall. Second in Google Scholar’s Computational Linguistics list.
- NAACL and EACL
- North American and European chapters of the ACLRegional conferences of the same association, with the same review system as ACL.
- AAAI
- AAAI Conference on Artificial IntelligenceA general AI conference held since 1980. Fourth in Google Scholar’s Artificial Intelligence list.
- CVPR
- IEEE/CVF Conference on Computer Vision and Pattern RecognitionThe main conference for AI that works with images and video. First in Google Scholar’s Computer Vision list. Its papers are cited more than those of any other AI venue.
- COLM
- Conference on Language ModelingA conference on language models that first met in 2024.
Terms
- Peer review
- Several researchers in the same area read a paper before it can be accepted. They score it and write comments. The authors do not choose them. Most submissions are turned down.
- Main conference
- The central track of a conference. It takes full-length papers and has the lowest acceptance rate. Most authors are PhD students, professors and industry researchers.
- Workshop
- A meeting on one research area, held at the conference. Leading researchers in the area organize it and peer review its papers. Accepted papers are presented at the conference.
- Findings
- Main track at ACL and EMNLP. Findings papers go through the same main-conference peer review and are published in the conference’s Findings volume.
- Oral and spotlight
- A distinction the conference gives to a small share of its accepted papers, the ones its reviewers rated highest. Their authors present them on stage, as a full talk (oral) or a featured short talk (spotlight).
- OpenReview
- The public website most AI conferences use to run peer review. Accepted papers are listed there with their venue, track and authors.
- arXiv
- A public archive run by Cornell University where researchers post papers. Anyone can post there. An arXiv entry shows only that a paper exists. The conference listing shows whether it was accepted.
Each of our students' accepted papers is labeled with its venue and track.
Begin Your Journey
The application takes 5 minutes and is reviewed on a rolling basis. We look for strong technical signal—projects, coursework, or competition results—and a genuine curiosity to do real research.
If admitted, you will join a structured pipeline with direct mentorship to take your work from ideation to top conference submission at venues like NeurIPS, ACL, and EMNLP.
