A fifth-year PhD student with no publications is a pressure cooker. The student is not asking for advice on research direction or how to improve a model. They are asking a far more pointed question: where can I go to get this done, not to get discovered? They name NeurIPS as part of an "elitist club" and they are right to feel that the game has changed for them. When you have been rejected from top venues despite good review scores, the problem is not your science. It is the system.
We should say plainly: this is not a story about a student failing. It is a story about how the publishing culture in AI has drifted away from its stated purpose. The student's work on efficient generative AI is not marginal. Efficiency is the bottleneck for real-world deployment, whether you are running vision models on a phone or compressing a diffusion model for a laptop. As we explore in Exploring Real-World Computer Vision: Deployments, Edge Models, and Current Challenges, the hard part is almost never the architecture; it is making it work under tight compute and latency constraints. That is exactly the kind of practical contribution that should be celebrated, not shunned. And if you look at tools like the Forrester Function, you see that the field has long valued mathematical clarity and reproducible methods over conference prestige. The student is already working in that spirit. The venues just have not caught up.
So what should they do? Stop aiming for the top and start aiming for the right fit. There are solid, peer-reviewed journals and mid-tier conferences that accept solid work without the circus of a 25% acceptance rate. The goal is to graduate, and graduating requires a published paper, not a keynote. We would tell that student to look for venues that value reproducibility and engineering rigor over novelty theater. Find a journal with a fast review cycle. Look for workshops attached to larger conferences that publish proceedings. And do not ignore the growing number of open-access venues that are building a reputation for fair, fast, and technical review. The work matters. The venue is just a receipt.
The deeper issue is that we have allowed prestige to substitute for progress. If you are doing efficient generative AI, your work is already about making AI accessible. That is a value we share in Verify Your AI's Understanding: A Simple Check for Tax Season, where the focus is on whether a model actually understands what it is doing, not on where it was published. The student should treat this as a signal to redefine success on their own terms. A degree is a credential, not a verdict on your potential. Publish where the work is read, not where it is judged. And when you do get that acceptance, do not look back. The people who rejected you were not protecting science. They were protecting a filter that no longer serves it.