iiinova — where your digital self keeps evolving

Reposted by Giorgio Gilestro

Stephen Turner @stephenturner.us · 26/09/2026

Personal essay, weekend edition. My relationship with AI is changing. 1.8k words, 8 min reading time. doi.org/10.59350/bh7...

doi.org

My relationship with AI is changing

When the facts change, I change my mind. What do you do, sir?

06615

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

I didn't have any specific example in mind and didn't know your service, sorry. I just think that when data is all we have to give, the natural instinct will be to protect it.

000

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

Part of me thinks the system will be self limiting in the west because it's incompatible with capitalism. It is however compatible with socialism so the world will fracture.

000

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

I agree that people realise about the real impact only when it slaps them in the face. The most recent example being the Fields medals' letter. I don't think this is something we can really avoid or resist. But at the same time I don't believe its growth is compatible with society as we know it.

210

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

No, I wrote: "I suspect irrelevance is a problem we will have forgotten in six months, in the same way we have basically forgotten about hallucinations." Irrelevance = slop

100

Reposted by Giorgio Gilestro

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

Biology is not safe. What I actually fear the most about biology is the funding squeeze: if on one hand you have AI promising thick and thin, funders will ask themselves is it even ethical to keep funding a small group studying the chemical composition of lizards saliva? (Not a random example!)

042

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

WDYM? I think most people around me use it but very few know how to use it.

100

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

Two steps ahead is where i am comfortable. :) There's massive need for regulation but regulation won't help you with this particular problem. I think you are paying too much attention to AI slop but slop will disappear soon (as hallucinations basically did). Slop is a very transient problem.

100

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

it has been proposed to interview submitters to check if they understand what their paper is about. I would not want to have a disease-curing paper rejected just because its "authors" have no understanding of what's going on. this may sound like a preposterous example now, but soon it won't.

100

Reposted by Giorgio Gilestro

paveltomancak.bsky.social @paveltomancak.bsky.social · 20/09/2026

Wow. Amazing, prescient vision of the future of academia by @giorgio.gilest.ro You don’t need to agree, but read and consider.

186

Giorgio Gilestro @giorgio.gilest.ro · 21/09/2026

042

Reposted by Giorgio Gilestro

Jean-Yves Tinevez @jytinevez.bsky.social · 21/09/2026

An incredible, eye opening piece

073

Giorgio Gilestro @giorgio.gilest.ro · 20/09/2026

My thinking is not justified by the AI slop though. I believe AI slop will no longer exist in a few years time and we will be flooded by a tsunami of technically sound incremental discoveries that only other AI will have time to digest. Then papers will cease to have value for people's careers.

2114

Giorgio Gilestro @giorgio.gilest.ro · 20/09/2026

Academia has seen inflation in the past 30 years but it is now entering Republic of Weimar territory, with value of knowledge moving from "paper to land". In all this, Open hardware is an act of armed resistance to Technofeudalism. giorgio.gilest.ro/the-weimar-o...

giorgio.gilest.ro

The Weimar of knowledge – la bottega

23318

Giorgio Gilestro @giorgio.gilest.ro · 16/09/2026

Writing a manuscript with this new generations of LLMs became an impossible task. I feel that moving away from RLHF towards RLVR increased problem-solving abilities at the cost of conceptual clarity and exposure. LLMs became brilliant engineers who cannot longer comunicate with humans.

020

Reposted by Giorgio Gilestro

Jose Jimenez @jjimenez.bsky.social · 15/09/2026

We are recruiting!! We are looking for an expert in the metabolic engineering of non model microorganisms to join our team working on plastic waste upcycling. For details please check the link below. www.imperial.ac.uk/jobs/search-...

imperial.ac.uk

Description

Please note that job descriptions are not exhaustive, and you may be asked to take on additional duties that align with the key responsibilities ment...

0312

Giorgio Gilestro @giorgio.gilest.ro · 13/09/2026

Here's an idea. Arxiv could peer review the papers and accept only those that pass peer review. Then even get an IF maybe! Insanity all around.

120

Giorgio Gilestro @giorgio.gilest.ro · 08/09/2026

On the Navier-Stokes drama. From this exchange, I find OpenAI to be rather accomodating and I would interpret the "why would you damage your career line" as "by publishing something that is rushed, AI slopped and half baked". The request to drop Alpoge is indeed weird but I can see the complcation.

000

Reposted by Giorgio Gilestro

Richard Sever @richardsever.bsky.social · 27/08/2026

$3.7 billion spent on APCs in 2025. That's ~1000X more than it costs to run bioRxiv/medRxiv. Worth thinking about as we consider how to fund infrastructure... arxiv.org/pdf/2608.16322

arxiv.org

813379

Giorgio Gilestro @giorgio.gilest.ro · 20/08/2026

The model builds on ideas that were put forward by many before, and was sharpened by suggestions and criticism of many colleagues. I had initially drafted this idea in a 2018 preprint. The reviewers suggested leaving that concept for a review. Here it comes, 8 years later. doi.org/10.32942/X22...

010

Giorgio Gilestro @giorgio.gilest.ro · 20/08/2026

I make predictions and hypotheses in the paper and explain how confident I am about each of them. I also apply the model to what we know about human sleep and, doing so, explain many apparent paradoxes linked to sleep length in humans.

110

Giorgio Gilestro @giorgio.gilest.ro · 20/08/2026

This is a review, and a positional paper proposing a new framework. This model has one major advantage: it explains all of the paradoxes of the field. It has a price, though. You may have to think about sleep in a different way and recognise the key role of evolution.

100

Giorgio Gilestro @giorgio.gilest.ro · 20/08/2026

Sleep is not one thing but a combination of processes with different elasticity. The vital component is life-saving and has zero elasticity; the useful component has some; the accessory component has the most, and lets sleep expand and contract in migratory birds, elephants - and us.

100

Giorgio Gilestro @giorgio.gilest.ro · 20/08/2026

Borrowing from economics, I introduce "sleep elasticity". The concept is Marshall's, 1890: essential goods like life-saving medicine and winter heating have low price elasticity, demand stays high as prices rise. Luxuries have high elasticity, demand drops sharply as prices rise.

100

Giorgio Gilestro @giorgio.gilest.ro · 20/08/2026

Variability is not just between species, but also within a single animal. Some birds, for instance, sleep 16 or more hours during the winter and then go to minutes a day during the migratory & breeding season. They forego sleep almost completely for weeks or months. How is that possible?

100

Giorgio Gilestro @giorgio.gilest.ro · 20/08/2026

All animals sleep. This is something we have rightly taken as evidence of its importance. However, they all sleep in different ways and for different amounts, from 2 to 20hrs a day, something that speaks against a unified cellular phenomenon. In biology, variability is always key to understanding.

100

Giorgio Gilestro @giorgio.gilest.ro · 20/08/2026

The reason for sleep to exist is still a mystery. "Why do I have to sleep" is not just a question every toddler asks, but, at some point, every curious mind. Many hypotheses have been put forward, some more fundamental than others. In this preprint, I try a different approach doi.org/10.32942/X22...

doi.org

Sleep elasticity: accessory, useful and vital components of a single state

293

Giorgio Gilestro @giorgio.gilest.ro · 18/08/2026

For that you would have to use a commercial anti plagiarism software.

101

Giorgio Gilestro @giorgio.gilest.ro · 17/08/2026

So do I, to the point I am explicitly proposing this to be a REF criterium.

020

Giorgio Gilestro @giorgio.gilest.ro · 17/08/2026

That's part of the problem though. Yes, everyone can do so and then compare against a virtuous baseline that is only imaginary. When people do perform that kind of exercises in the way it should be done (a large, unbiased screening) it turns out 1 in 2 Doctoral theses have issues with plagiarism.

100

Giorgio Gilestro @giorgio.gilest.ro · 17/08/2026

Nowadays there are instruments for detecting plagiarism automatically, cheaply, without effort. As I wrote in that piece, I would be totally in favour for those to be applied to every professor and every student, at Cambridge and elsewhere. Let's build a baseline before throwing stones.

100

Giorgio Gilestro @giorgio.gilest.ro · 16/08/2026

Four claims about Jason Arday spread like wildfire in the days around his death: that he was unqualified, that he plagiarised his thesis, that he was hired as a poster boy, that he lied his way to the top. All four are checkable, so I checked them. giorgio.gilest.ro/four-lies-ar...

giorgio.gilest.ro

Four lies about Jason Arday that spread too easily – la bottega

02310

Giorgio Gilestro @giorgio.gilest.ro · 16/08/2026

It means that 4.2% of words belong to sentences that are not unique. Anything below 10% is usually ignored because considered physiological. Verbatim copying long sentences is frowned upon but extremely common. In other words, there is nothing worth this level of scrutiny in that work.

100

Giorgio Gilestro @giorgio.gilest.ro · 15/08/2026

When LJMU investigate the case, in March 2026, it concluded the overlap to be "well within the accepted range of compliance with the academic standards of the time". That should have been the reference for journalists, not an accusatory blogpost raised by someone with little or none academic credit.

040

Giorgio Gilestro @giorgio.gilest.ro · 15/08/2026

I did find some verbatim sentences but it would be naive to expect none: studies that measured plagiarism, e.g. Ison 2014, estimate a mean similarity index of 15%. I found 4% in a 1-to-1 match, not far from what reported by Arday's solicitors in their letter to THE: 2-7% against all literature.

120

Giorgio Gilestro @giorgio.gilest.ro · 15/08/2026

It was alleged that Jason Arday was an academic fraud mainly based on accusation of plagiarism initially moved by Nathan Cofnas. I have compared Arday's thesis with the one he's been accused of copying and found a bare 4.2% similarity.

299

Reposted by Giorgio Gilestro

Gregory Kohn @kohngregory.bsky.social · 15/08/2026

In the wake of Arday's death, Nathan Cofnas (who led the X mob against him) states his motivation. Beyond whistleblowing, he states that after a 'revolution' removes professors like Arday, it will open opportunities for him to become 'head of the department of eugenics and race science at Harvard'.

51434461311

Reposted by Giorgio Gilestro

Richard Sever @richardsever.bsky.social · 06/07/2026

“move from preprint to peer-reviewed publication leaves the central claims of most abstracts intact, indicating preprints are a reliable source…papers that were never posted as preprints were retracted at roughly twice the rate of those that were…” www.biorxiv.org/content/10.6...

biorxiv.org

Tracking claim changes from preprint to publication across 72,644 biomedical studies using large language models

Preprints now disseminate a large share of biomedical research before peer review. Because they have not yet passed peer review, some scientists regard preprint claims as unverified or potentially unr...

611158

Reposted by Giorgio Gilestro

Julie Seibt @jseibt.bsky.social · 19/06/2026

Heading to FENS 2026 and curious about the latest thinking on sleep function(s)? Join us for a networking event exploring sleep across species, with amazing speakers Peter Meerlo (Netherlands), Meg Crofoot (Germany), @giorgio.gilest.ro (UK) More info: fens2026.abstractserver.com/program/#/de...

fens2026.abstractserver.com

Program

021

Reposted by Giorgio Gilestro

Julie Seibt @jseibt.bsky.social · 05/06/2026

Applications are now open for the inaugural CAJAL Advanced Neuroscience Training Course on Neurobiology of Sleep 😴🧠!!! 📅 Dates: 19 November – 8 December 2026 ⏳ Application Deadline: 3 July 🔗 Information and application: cajal-training.org/on-site/neur...

083

Giorgio Gilestro @giorgio.gilest.ro · 19/05/2026

And I forgot to mention the "island of strangers" pitch. The guy spent months courting reform voters, and now he wonders why he lost labour voters. Not quite the sharpest tool in the box.

040

Giorgio Gilestro @giorgio.gilest.ro · 19/05/2026

I cannot say precisely what the sign actually read because I may get in trouble for just repeating that because the government appealed against the court (they really had to double down didn't they?). Hiding the situation and gaslightinging the public is why labour is where it is.

210

Giorgio Gilestro @giorgio.gilest.ro · 19/05/2026

I'd be happy with anyone who is not right wing tbh. No bashing on migrants, no changes to ILR policies, not arresting elderly people for holding a sign that says Palestine. That would be a good start. Taxing some billionaire rather than cutting winter fuel subsidies may help too.

150

Reposted by Giorgio Gilestro

Harmit Singh Malik @harmitmalik.bsky.social · 23/04/2026

Beautiful creative work and a great 🧵 Kudos to all involved in this nice work 🧪

0275

Reposted by Giorgio Gilestro

Junior European Drosophila Investigators @fly-jedi.bsky.social · 15/03/2026

Calling all #Drosophila #NewPI #researchers! A few spots remain for the upcoming JEDI Conference - and you won’t want to miss this! Every attendee gives a talk, making the exchange dynamic, personal, and highly interactive - all at a stunning venue that inspires collaboration and creativity.

View of the historic hotel Pousada Castelo in Palmela.

02621

Reposted by Giorgio Gilestro

Carlos Ribeiro @ribeirocarlitos.bsky.social · 31/03/2026

Immense respect for handling this situation with this level of integrity. Mistakes happen. Science is a process.

87611