Jonathan Richard Schwarz flags how PhD spots at the new Thomson Reuters-Imperial Frontier AI Research Lab now routinely go to candidates who already list multiple first-author papers plus orals at NeurIPS and similar venues, a threshold he says sits noticeably higher than five or ten years ago and can sideline other capable applicants. He later clarified that several accepted students meet this mark rather than every single one.
PhD apps in 2026 are wild: Every successful applicant to our new lab @imperialcollege has multiple 1st-author papers, incl. several orals at NeurIPS, etc.
Feel deeply for students who'd have earned a spot 5–10 years ago — same talent & drive, just a different moment in time.
David Pfau, nearly twenty years into machine learning research, replied that he has never presented an oral at a top venue, underscoring how research careers can diverge even among established contributors.
The reported bar applies only to the initial cohort of this one newly funded lab; exact applicant numbers, acceptance rates, and whether the pattern holds for other Imperial programs or UK labs remain unstated.
Many users criticized AI labs requiring multiple NeurIPS orals and first-author papers for PhD applicants as unrealistic credential inflation that discourages talent and encourages sloppy work.
No Digg Deeper questions have been answered for this story yet.
PhD apps in 2026 are wild: Every successful applicant to our new lab @tri_fair_lab has multiple 1st-author papers, incl. several orals at NeurIPS, etc.
Feel deeply for students who'd have earned a spot 5–10 years ago — same talent & drive, just a different moment in time.
@schwarzjn_ @tri_fair_lab Don't you think you're part of the problem? (I don't mean it in an aggressive or offensive way, sorry)
Like, there's a master's student or third year undergrad who's going to read your tweet, take it as the standard, and start tossing as many submissions to NeurIPS as they can.
@schwarzjn_ @tri_fair_lab Half the reason there are so many sloppy papers submitted to every top tier conf. is because labs at top schools keep signaling that a bunch of accepted papers is a better way to get accepted rather than, say, one decent paper plus strong letters, grades, experience, etc.
@schwarzjn_ @tri_fair_lab Most of them go to some lab because they were told they need a paper very early on. They realistically have no idea whtas going on at a deep level but they’re at a good enough school to find a group about to publish so they can attach their name. Information asymmetry.
@benno_krojer @tri_fair_lab Let me clarify: Several of the accepted students have orals at top venues, not all of them
@schwarzjn_ @tri_fair_lab Each one of them has more than one oral at neurips? That sounds crazy
@schwarzjn_ @tri_fair_lab how is it possible? are they master's student who published research during university? independently?
@schwarzjn_ @benno_krojer @tri_fair_lab I've literally never had an oral at a top venue in my almost two decade long career in machine learning research.
@benno_krojer @tri_fair_lab Let me clarify: Several of the accepted students have orals at top venues, not all of them
@schwarzjn_ @tri_fair_lab Same talent and drive? All these 10 papers in 2 years people are often frauds or at best leveraging their network. Leaving aside the physical impossiblity, I wonder if they even know the fundamentals of the field their papers claim they are experts in.
@schwarzjn_ @tri_fair_lab I agree. I have been trying to land a phd since 2024. The amount of people that have first author is has increased ten folds every year. Do you have any suggestions? Maybe something that you think would help standing out? @schwarzjn_
@schwarzjn_ @tri_fair_lab Do you think candidate quality is going up or is it credential / achievement inflation similar to the situation in the high-school -> college pipeline?
@ShcChy @schwarzjn_ @tri_fair_lab This. Also you can have all of the aforementioned, but if you have no slop paper you can say goodbye to your PhD.
@schwarzjn_ @tri_fair_lab Orals??
@schwarzjn_ @tri_fair_lab Multiple 1st author papers on NeurIPS? With just a master's degree?? I feel so left behind
@schwarzjn_ @tri_fair_lab Yea ig its time to pack my bags and call it quits for a few years