iiinova — where your digital self keeps evolving

Nicolai Friis @improbabilitydrive.bsky.social · 28/07/2026

I fully agree with this. How are journals/editors/reviewers going to handle this, e.g., at @quantum-journal.bsky.social? Previously, we were worried about mass-produced AI slop flooding submission, but now it seems we have to find a way to assess the significance of serious work done with/by AI.

000

Nicolai Friis @improbabilitydrive.bsky.social · 27/07/2026

As a bigger-picture question (beyond my personal wishes on how to do my job), I do wonder how this development will impact the ability of humans/society in general to train their skills in reasoning, logic, maths, etc.

050

Nicolai Friis @improbabilitydrive.bsky.social · 27/07/2026

Maybe no labels like "good" or "bad" can be attached to this, it merely comes down to personal perspective, but if the "social contract" becomes that being a scientist necessitates the use of AI to prove things, at least I would be disincentivized to keep doing this job

Nicolai Friis @improbabilitydrive.bsky.social · 27/07/2026

This is certainly a valid perspective. And one could argue that any job will come along with expectations as to how to perform it efficiently. And yes, maybe it is selfish, but I think a job is also self-fulfillment, and enjoying a job is a big part in choosing it in the first place.

Nicolai Friis @improbabilitydrive.bsky.social · 27/07/2026

I see your point, fair enough, surely AI tools are a game changer. I am just not sure I want to be the metaphorical truck driver though: a big part of what got me interested in this profession was the creative aspect/challenge that comes with trying to prove things.

Nicolai Friis @improbabilitydrive.bsky.social · 27/07/2026

I agree that one has to take into account the pressures of getting jobs/grants to be able to keep doing science, I just wonder if I would choose this career path in the first place if the premise is that one has to outsource the creative aspects to AI in order to get/do this job

240

Nicolai Friis @improbabilitydrive.bsky.social · 22/07/2026

Why would I want to pay for AI to do the thing that I enjoy about being a scientist? I don't do science to be competitive or to beat someone to the punch. I genuinely don't understand why people would want to have AI do all the creative aspects of their work.

680

Nicolai Friis @improbabilitydrive.bsky.social · 19/07/2026

Ist "Horrorfilm" ein Euphemismus für das WM-Finale? Könnte das durchaus nachvollziehen.

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Preprint: It's not strictly necessary to add @arxiv.bsky.social information for published articles, and many publishers will remove it in published versions, but I think it is good practice to add arXiv numbers and remind people to put things on the arxiv to make research publicly available. (16/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Doi: Everything that is published online these days (and many things that have been published a while ago) is assigned a digital object identifier (doi), which means that the document can be found online on a permanent url doi.org. Best use that in the url = {} field. (15/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Volume, Pages, Year: Make sure they are there. That's it. Page numbers these days are often not actual page but article numbers, but they still go in the pages = {...} field. Be careful of publishers using ascii hyphens for page ranges, your compiler will not like that. Use {\textendash}. (14/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Journal field: If your style file can switch between abbreviated and non-abbreviated titles, put the full title, otherwise pick one and be consistent. Make sure to use standard (iso4) abbreviations. (13/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Title: if there are capitalized words in the title (e.g., "Hilbert" or the title has every word capitalized as in the titles of PRLs), then these capitals will not be printed unless one wraps those letters or the entire title in exta {...}: use title = {{This is the Title of the Paper}} (12/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

I would also encourage using full first names in addition to last names. This gives more visibility to underrepresented groups and also makes it easier to distinguish authors with the same last name. Saving space on a printed page really is not a concern anymore today. (11/16)

130

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Authors: Use the format author = {Lastname, Firstname and Lastname, Firstname and … } so that bibtex can properly resolve multiple first or last names and suffixes correctly. Use tex commands for accents/special characters, utf8 input encoding may not work with some sty files. (10/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Also be careful with importing/copying bibtex entries from publisher websites, they often make basic mistakes. proper nouns in titles are not protected, DOIs and URLs are misidentified/exchanged, URLs lead to publisher websites, page numbers missing etc. So what to look out for? (9/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

...it will be filled with random imaginary author lists, fake journal references for preprints, fake preprint numbers, fake DOIs, and other fun AI hallucinations. Once this has happened, you will have no choice but to cross-check every single entry, line by line, name by name, digit by digit. (8/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Now comes the advice that goes along with setting up a bib file: The most important thing first, and I am not sure who needs to hear this, but DO NOT LET LLMs TOUCH YOUR BIB FILE! Once the not so intelligent AI has gotten its grubby digital fingers on your carefully curated .bib file... (7/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Step 4: Along with your template and .bib file, you want to chose/create/modify one or more style files for bibtex. Make sure the style files call the fields you use in the bib file, or vice versa, that your bib file contains the fields the style files call. (6/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Step 3: Set up a .bib database for bibtex. This will be the centre piece for your bibliography. I use one such file, I add entries by hand, if you check each entry once, you can be sure it is correct and continue to use and updated the same .bib file for every project. (5/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Step 2: Create working templates for the types of documents you usually handle, articles, theses, etc., make sure they compile on your machine. You can use those also whenever you do start a project on Overleaf. Almost anything will compile there, but not necessarily the other way around. (4/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Sure, you can use Overleaf. I use it a lot. But I have local copies of all my papers. It often happens that there is no internet connection, some DDOS attack, or other fun reasons why one cannot access Overleaf. You can of course simply hope this doesn't happen the day before a deadline. (3/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

Step 1: Use (La)TeX with Bibtex. On. Your. Computer. Yes, this means installing a Tex distribution on your machine. It is easy and free, find a decent distribution, like MikTex and a compiler, I like WinEdit, but use whatever you like/have access to. "I just use Overleaf", you will say. (2/16)

Nicolai Friis @improbabilitydrive.bsky.social · 14/07/2026

One thing that I really have a hard time understanding is why it seems to be so difficult for researchers (at least in my field/experience) to get one of the most basic things in scientific writing correctly: references. So here is my (unsolicited) advice on how to sort out your references: (1/16)

141

Nicolai Friis @improbabilitydrive.bsky.social · 12/07/2026

Yesterday someone called me out for only having joined Instagram to drop the occasional Pratchett quote without explanation. I mean, it isn't not true. But: It's not worth doing something unless someone, somewhere, would much rather you weren't doing it.

020

Nicolai Friis @improbabilitydrive.bsky.social · 12/07/2026

Coincidentally (or is it?!) I tend to find Cthulhu on my toast quite regularly. Or in my spaghetti. Sometimes even in my scientific papers. 🐙

000

Nicolai Friis @improbabilitydrive.bsky.social · 06/07/2026

So, if it seems like things are at times looking dreary for you as a mentor or mentee, supervisor or student, hang in there, believe in yourself and believe in your students, be patient, be kind, be understanding, everyone has bad stretches, but things will be looking up again! (8/8)

Nicolai Friis @improbabilitydrive.bsky.social · 06/07/2026

In the meantime, my student Tomasz Andrzejewski and Esprit Fellow Phila Rembold managed to put the finishing touches on a longstanding project and I am very happy to say that we just uploaded the resulting paper on creating genuine multipartite entanglement between logical qubits to the arXiv. (7/8)

130

Nicolai Friis @improbabilitydrive.bsky.social · 06/07/2026

This was followed by another brilliant PhD defense by Klára Baksová, who (unfortunately for me) will soon be leaving the group to start a postdoc in Prague, I am sure she will succeed there! (6/8)

140

Nicolai Friis @improbabilitydrive.bsky.social · 06/07/2026

Then I had the pleasure to attend the 2026 Seefeld Quantum Information Workshop, where my student Ida Mishra enthusiastically presented a poster on her upcoming paper on high-dimensional entanglement and APART-MINT Fellow Simon Morelli advertised his work on sequential parameter estimation. (5/8)

150

Nicolai Friis @improbabilitydrive.bsky.social · 06/07/2026

On 17 June, my (by now I should say former) student Nicky Kai Hong Li defended his PhD with flying colours, and I am very happy that he will be staying on for a postdoc with a Croucher Fellowship! (4/8)

140

Nicolai Friis @improbabilitydrive.bsky.social · 06/07/2026

So I consider it an imperative to be there for my students when there seems to be no light at the end of the tunnel. The flip side: It is then with even greater joy that one can witness the moments where one sees one’s students succeed, and recently I had this privilege on several occasions... (3/8)

Nicolai Friis @improbabilitydrive.bsky.social · 06/07/2026

As a supervisor, one feels the pains and pressures of the talented people that one is trying guide through difficult patches, not least because more experienced academics in supervising positions have themselves experienced directly how tough, competitive, and sometimes cruel academia can be. (2/8)

Nicolai Friis @improbabilitydrive.bsky.social · 06/07/2026

Bear with me, this will end on a high note. At times I feel that the workload and responsibility that comes along with supervision can be a lot, often because besides the actual work, academia as well as life outside of academia can be challenging emotionally, both for mentors and mentees. (1/8)

150

Nicolai Friis @improbabilitydrive.bsky.social · 19/06/2026

Wait a second ... ISTA's lecture hall is called the "Raiffeisen Lecture Hall"? Since when is it good practice to name lecture halls at universities after sponsors? In particular, sponsors still in business w. countries starting wars of aggression, committing war crimes & political assassinations?

Nicolai Friis @improbabilitydrive.bsky.social · 18/06/2026

What a brilliant day, three consecutive defenses: It was an honor to serve on the PhD committee of Philipp Sohr (middle), and then very proud to see my student Nicky Kai Hong Li (second from right) defend valiantly, before leaning back as Marcus' (left) student Iraj (second from left) took the stage

031

Nicolai Friis @improbabilitydrive.bsky.social · 04/06/2026

An answer many parents might understand better than what we have actually been doing all these years. They might (have) even be(en) the ones to hand over the champagne and cigar.

020

Nicolai Friis @improbabilitydrive.bsky.social · 05/05/2026

Kürzeste exakte Beschreibung Österreichs, die mir bisher untergekommen ist