iiinova — where your digital self keeps evolving

Simon Cleary @simoncleary.bsky.social · 04/10/2026

Winner furiously trying to reverse engineer raw data using ChatGPT. Judges trying to meet but keep getting distracted by tiktok dramas about humanoid fruits, tung tung tung sahur dance videos, Rastawartz...

020

Reposted by Simon Cleary

Dale Moulding @dalemoulding.bsky.social · 01/10/2026

Having seen the news: www.bbc.co.uk/news/article... thought I'd share a cilia movie captured at UCL GOS ICH with Eriomina Shahaj & Vanesa Papastavrou using simple DIC on a Nikon Ti2 with DIC at 60x1.4. 500 fps here slowed down 20 times. Processed with Nikon's denoiseAI and an unsharp mask in Fiji.

88622

Simon Cleary @simoncleary.bsky.social · 01/10/2026

It was good that Nature News did comment on the winner's unwillingness to show non-edited images. It is telling and was missed in the pieces from The Scientist and BBC News.

021

Simon Cleary @simoncleary.bsky.social · 01/10/2026

This stood out in the Nature news coverage. I don't think that we will ever get to see the underlying data.

141

Simon Cleary @simoncleary.bsky.social · 01/10/2026

The SynthID watermark is a smoking gun. It means Google or ChatGPT AI image generating software was used. There is no appropriate reason for these tools to be used in microscopy.

151

Simon Cleary @simoncleary.bsky.social · 01/10/2026

Rule: no AI-generated images. The winner confessed to AI-generating all details in lower half of image! SynthID watermark present, so an inappropriate Google/chatGPT AI generation tool was used. Movie also biologically implausible - we are told it is one cell but it is over 50 microns wide!

141

Reposted by Simon Cleary

Academy of Medical Sciences @acmedsci.bsky.social · 29/09/2026

Applications are open for our new fellowship programme with @lifearc.bsky.social 🚀 The scheme supports clinician scientists to build the skills, networks and experience needed to turn innovations in rare diseases into real-world impact. See eligibility and apply: bit.ly/4z1bUNQ

Professor Sir Andrew Morris, President of the Academy of Medical Sciences, next to a quote that reads: "With LifeArc, this fellowship gives clinician scientists in rare disease research the funding, mentoring and connections to put their research to use for the patients who are still waiting for it."

035

Reposted by Simon Cleary

Jez Carlton @jezcarlton.bsky.social · 28/09/2026

Hey! Submit your papers, and spread the word, this will be great! Thanks for inviting us @jcellsci.bsky.social and great to work with @pihanson.bsky.social on this issue!

02324

Simon Cleary @simoncleary.bsky.social · 26/09/2026

A bit worried that this statement from Nikon seems to be referring to the post-processing movie as the source video

250

Simon Cleary @simoncleary.bsky.social · 25/09/2026

There is a clear distinction between processing images to remove noise + spillover to better see real things (fine) and generating fake structures and time points (not fair!)

020

Simon Cleary @simoncleary.bsky.social · 25/09/2026

Unusual that the winner suggests that we ask how AI was used but does not provide answers... or the source data used to generate the movie.

010

Simon Cleary @simoncleary.bsky.social · 25/09/2026

Update on the Nikon Small World in Motion competition - Nikon now calling the AI-generated winning movie a 'rendition' and investigating:

100

Reposted by Simon Cleary

Valentin Dunsing-Eichenauer @dunsingvalentin.bsky.social · 25/09/2026

#FluorescenceFriday: Given the recently grown interest in ciliary beating, here is a movie of beating cilia in a human airway model that I captured last week with lattice light-sheet microscopy. No AI „processing“, this is the raw data! Frame rate is ca. 170fps, displayed here at 25fps.

311230

Reposted by Simon Cleary

Robin Bisson @robinbisson.bsky.social · 25/09/2026

Update on the "AI-based triage" which is getting lots of (negative) attention

In recent days, there have also been several reports on social media of applicants seemingly being rejected after “AI-based triage”. The funding, while originating from UKRI, was devolved to the Crane network, which is funded by the Engineering and Physical Sciences Research Council to strengthen the UK’s cybersecurity research ecosystem.

UKRI confirmed it did not carry out the AI triage process. A spokesperson said: “UKRI policy does not allow the use of generative AI to conduct application assessment. We are following up with the Crane network to review this situation in detail and ensure compliance with our assessment conditions.”

31812

Simon Cleary @simoncleary.bsky.social · 25/09/2026

I would prefer a lottery system to a potentially biased AI-triage. Would also have lower costs for the funder.

060

Simon Cleary @simoncleary.bsky.social · 24/09/2026

Sounds sensible as Nikon comms do not seem to understand. The winner's responses have made people even angrier, but I think it is important to remember that this is a young scientist who also doesn't seem to understand the upset, and pile-ons can be harmful. Wrote this on Linkedin:

Screenshot of Linkedin comment response to Ning Xu - "Ning - I want to add that the greyscale images in your preprint look excellent and you shouldn't feel discouraged from research by how other scientists have reacted to your submission to this contest. People are upset about use of AI to generate images here because doing this gives the false and potentially harmful impressions that airways look like this (images from this competition get used a lot in education and science communication) and that we can already image them like this (it is important to be open about limitations of current approaches so that research funders understand the need to continue to support research). "

262

Simon Cleary @simoncleary.bsky.social · 24/09/2026

@healthcare.nikon.com - @cilialab.bsky.social is right, and promoting AI-generated images in this way sends the message that people do not need to buy the nice microscopes that you sell

190

Reposted by Simon Cleary

Simon Cleary @simoncleary.bsky.social · 24/09/2026

Even if it were ok to add AI-generated details in the lower half of the movie, what we are now told is a section of one ciliated cell is way too wide to be based on real images of one cell.

4188

Simon Cleary @simoncleary.bsky.social · 24/09/2026

Even if it were ok to add AI-generated details in the lower half of the movie, what we are now told is a section of one ciliated cell is way too wide to be based on real images of one cell.

4188

Simon Cleary @simoncleary.bsky.social · 24/09/2026

...and I think that's important. The bottom half of the image is now stated as being faked, or inappropriately extrapolated from phase contrast details. If some of the cilia or some time points are also not from real signal, the images should definitely not be in the competition

020

Simon Cleary @simoncleary.bsky.social · 24/09/2026

I agree, the winning image sequence appears too large to be a single cell

130

Simon Cleary @simoncleary.bsky.social · 24/09/2026

I wrote an email to the address on the Nikon Small World website on Monday expressing my concerns and my appreciation of the contest. No response yet.

010

Simon Cleary @simoncleary.bsky.social · 23/09/2026

Why would a company that sells high-end cameras and microscopes want to give the impression that we could just use computers to dream up pretty pictures for us?

042

Reposted by Simon Cleary

Daniel Fisch @danielhfisch.bsky.social · 23/09/2026

Out now @nature.com! We report a new means of cell communication: receptor-free signal transduction. TLRs engage MyD88 for minutes, yet myddosomes signal for hours. TRAM releases myddosomes from TLRs, letting them mature and sustain signaling in the cytosol. rdcu.be/jGdKxErXyAsa Details below:

43214

Simon Cleary @simoncleary.bsky.social · 23/09/2026

There is a difference between applying LUTs and making efforts to reduce shot noise to enable viewers to interpret the true positions of real photons, and inappropriately applying AI tools to produce fake structures

110

Simon Cleary @simoncleary.bsky.social · 23/09/2026

So, it seems the winner was open about AI use but the judges did not recognise that AI has been inappropriately applied in this instance. People editing comms knew that AI enhancement/generation would be controversial so they removed mention of it in the piece on the winner.

170

Simon Cleary @simoncleary.bsky.social · 23/09/2026

In the original version of this piece the person/people writing for the Nikon competition also removed a mention of AI by the winner:

Original paragraph reads:
'His advice to other researchers considering entering the competition is straightforward: “Take care of the sample, use as little light as possible when working with living cells, and most importantly, keep recording. The best moment usually appears right after you decide to stop.”'

New paragraph has added section on AI: 'His advice to other researchers considering entering the competition is straightforward: “Take care of the sample, use as little light as possible when working with living cells, and embrace AI to reveal the hidden beauty in your data. Most importantly, keep recording. The best moment usually appears right after you decide to stop.”'

2185

Simon Cleary @simoncleary.bsky.social · 22/09/2026

These tools are easy to use but produce hallucination artifacts when applied to images different from those in the training dataset. In the case of the winning sequence, these are totally fake, mis-scaled details resembling features of confocal images of stained membranes, nuclei, and mitochondria

053

Simon Cleary @simoncleary.bsky.social · 22/09/2026

Nikon seem to have admitted awarding an AI-generated video 1st place in the Small World in Motion contest, which has a rule forbidding AI-generated entries:

1157

Simon Cleary @simoncleary.bsky.social · 22/09/2026

Info on the gecko hand image from Grigorii Timin and Michel Milinkovitch here: doi.org/10.1016/j.is... There is value in wonder, in showing data to new audiences and in being motivated to produce exceptional work. I hope Nikon continue to support competitions, but with some extra methods checks!

020

Simon Cleary @simoncleary.bsky.social · 22/09/2026

Some upset regarding the Nikon Small World in Motion winner this year so wanted to highlight that the beautiful image that won the Small World competition in 2022 was the initial inspiration for our project:

Embryonic hand of a Madagascar giant day gecko (Phelsuma grandis) stained for collagen and beta tubulin, from Grigorii Timin and Michel Milinkovitch https://doi.org/10.1016/j.isci.2023.106452

150

Simon Cleary @simoncleary.bsky.social · 21/09/2026

Grateful for any feedback or questions. Thanks to @acmedsci.bsky.social @balrcommunity.bsky.social @britpharmsoc.bsky.social @bioimaginguk.bsky.social for funding that enabled this work - a result of the summer projects of talented undergraduate students Amy Rivers and Mahhum Saqib!

Co-authors Amy Rivers (left) and Mahhum Saqib (right) in the lab.

042

Simon Cleary @simoncleary.bsky.social · 21/09/2026

First preprint from the lab! We found that fluorescence imaging of fibrillar collagen with fast green FCF is useful for imaging pulmonary fibrosis: doi.org/10.64898/202...

Two images showing fast green FCF staining of fibrillar collagen in magenta. An image of a fibrotic sample on the right shows higher intensity and prevalence of staining compared to a healthy control sample on the left.

2193

Simon Cleary @simoncleary.bsky.social · 18/09/2026

Good to see some active discussion on Bluesky though! Sort of a pile-on but as interesting and probably kinder than peak Twitter.

130

Simon Cleary @simoncleary.bsky.social · 18/09/2026

The winning image sequence is very different from images in a recent preprint from the winner: bsky.app/profile/simo...

010

Simon Cleary @simoncleary.bsky.social · 18/09/2026

You can see the nice (label-free) images that this microscope modality produces in this preprint: doi.org/10.48550/arX... Not sure how you would get from this to the prize-winning image (with very long anenome-like cilia) without genAI shenanigans...

161

Simon Cleary @simoncleary.bsky.social · 18/09/2026

No idea, this info should be provided. I wonder if pseudocolouring is allowed (arbitrarily pseudocoloured SEM images have won Wellcome image awards).

000

Simon Cleary @simoncleary.bsky.social · 18/09/2026

Would love more info on the colours and modality. A moving plane of focus might explain the 'shifting' nuclei and appearance of cilia.

210

Simon Cleary @simoncleary.bsky.social · 18/09/2026

Nikon's tool 'Denoise.ai' can produce strange artifacts when applied to images of unstable (live tissue) preparations. They should probably add a big warning message on this.

210

Simon Cleary @simoncleary.bsky.social · 18/09/2026

@cilialab.bsky.social I have seen non-PCD human cilia in primary explants and they are amazing, but this image seems too good to be true. Do you think it could be real?

130

Reposted by Simon Cleary

Waggoner Lab @labwaggoner.bsky.social · 05/08/2026

Neutrophil NADPH oxidase breaks the inflammatory IL-1β/IL-17A circuit to enhance pathogen clearance during respiratory virus infections @natcomms.nature.com @kingscollegelondon.bsky.social www.nature.com/articles/s41...

196