Lightnews — Scholar-powered news

Adam Kleczkowski @aklecz.bsky.social · 3d

As usual, a great pleasure to be at the @planthealthscot.bsky.social annual conference.

A poster on Ips typographus, and a talk on #AI in #planthealth #PlantScience

1

Adam Kleczkowski @aklecz.bsky.social · 10d

New (very short) post on my blog on AI:

statisticallyinsignificant.blog/2025/09/30/a...

AI – The Good, the Bad and the Ugly: Part 2

It is going to be a very short post today. We increasingly see that companies like LinkedIn, Facebook, and Google are starting to use our personal data and content we create to train generative AI …

statisticallyinsignificant.blog

3

Adam Kleczkowski @aklecz.bsky.social · 16d

What a feeling - just submitted to @ukri.org a large £3m grant application with 5 universities/collaborators.

It is my third application this year, first as a PI, total value £7m.

@unistrathclyde.bsky.social

1

Adam Kleczkowski @aklecz.bsky.social · 19d

It's this time of the year. It will be a delicious breakfast tomorrow.

#fungi #mushrooms

Mushrooms, one held by hand, more in a basket

2

Adam Kleczkowski @aklecz.bsky.social · 23d

share.google/4ejVNRc9OuJZ...

Learning the Quadratic Equation (UK) Doodle - Google Doodles

Learn more about the creation of Learning the Quadratic Equation (UK) Doodle and discover the story behind the unique artwork.

share.google

1

Adam Kleczkowski @aklecz.bsky.social · Sep 8

theconversation.com/we-risk-a-de...

An article describing how AI is changing the scientific publication landscape. A combination of AI writing and predatory publishing leads to a deluge of "resmearch" - or rather "research slop".

We risk a deluge of AI-written ‘science’ pushing corporate interests – here’s what to do about it

We have to tighten up on what research policymakers rely on, and also the inner workings of peer review.

theconversation.com

1

Adam Kleczkowski @aklecz.bsky.social · Sep 7

Carcassonne! Our favourite family game!

Bought Dominion and will be reading it when travelling to Nottingham for the BSPP conference next week.

1 4

Reposted by Adam Kleczkowski

Andrew Heiss @andrew.heiss.phd · Sep 6

oh noooo

compared to last year’s results.” However, I take issue with the author’s subsequent interpretation of the most recent test scores which reflects poorly on his/her understanding of basic statistics. The differences in proficiency rates cited in throughout the article are overwhelmingly NOT STATISTICALLY SIGNIFICANT.

Using a standard p-value of 0.05, any improvement in scores must show at least a gain of five percentage points in order to be considered as statistically significant. So, for example, even the slight decline in 4th grade ISAT scores in Idaho (2024 67.1%, 2025 64.6) is not significant. Similarly, the following assertion made by the author of the article is incorrect and misleading.

“Students scoring in the advanced range of the proficient category saw a slight increase, with 23.2% scoring advanced in 2024 compared to 23.6% in 2025.” In order to have been statistically significant, 28.2% of students must have scored proficient in the language arts subtest.

20 64 410

Adam Kleczkowski @aklecz.bsky.social · Sep 7

An autumn in Scotland – Chopin, 1848 bylines.scot/society/an-a...

An autumn in Scotland – Chopin, 1848

A poetic ode to Chopin and his Scottish benefactor Jane Stirling

bylines.scot

1

Reposted by Adam Kleczkowski

Matthew Facciani @matthewfacciani.bsky.social · Sep 5

Here are some of the false or misleading claims RFK Jr. made during yesterday’s hearing:
ca.news.yahoo.com/fact-checkin...

Fact-checking RFK Jr.'s testimony during Thursday's hearing

The HHS secretary repeated several false and misleading claims.

ca.news.yahoo.com

1 11 41

Adam Kleczkowski @aklecz.bsky.social · Sep 6

Science Under Siege: A Talk With Peter Hotez and Michael Mann
open.substack.com/pub/paulkrug...

Science Under Siege: A Talk With Peter Hotez and Michael Mann

Lysenko comes to America

open.substack.com

2

Reposted by Adam Kleczkowski

Raywat Deonandan @deonandan.bsky.social · Sep 5

Well, you're the FDA Commissioner, so tell us. What do the data say?

No? Here, I'll help you.

Thread [1/5 ]

CDC: deaths within 42 days of mRNA vaccination were below background (expected) rates in every age group, including ages 6 months–17 years

www.cdc.gov/acip/downloa...

11 170 510

Adam Kleczkowski @aklecz.bsky.social · Sep 6

You will be surprised how easily and quickly it could happen, see 1930s Germany or Soviet Union.

Reposted by Adam Kleczkowski

Emilio Vilanova @evilanova.bsky.social · Sep 4

www.eldestapeweb.com/sociedad/cie... #Treemortality Forests in Patagonia are experiencing high levels of tree mortality most likely linked to increasing temperatures and severity of droughts

Registran episodios de mortalidad masiva de árboles en bosques patagónicos

Se formó una red de más de un centenar de investigadores y voluntarios para monitorear el crecimiento de árboles en todo el país

www.eldestapeweb.com

1 10

Adam Kleczkowski @aklecz.bsky.social · Sep 4

On genAI bubble:

"Once, men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them."

F. Herbert, Dune

1

Adam Kleczkowski @aklecz.bsky.social · Sep 3

Bovine Tb was rampant then and other bacterial diseases.

Souring milk was an alternative.

bTB is back, at least in the UK.

3

Adam Kleczkowski @aklecz.bsky.social · Sep 2

One week until the @bspp.bsky.social #PPATH25 conference:

www.bspp.org.uk/conference-i...

See you all there, and please come to our special session on the Artificial Intelligence revolution in plant pathology research!

#PlantScience

Conference Information Page – Plant Pathology 2025 (#PPATH2025) and Early Careers Plant Pathology 2025 (#ECPP2025) – BSPP – The British Society for Plant Pathology

www.bspp.org.uk

3 3

Adam Kleczkowski @aklecz.bsky.social · Sep 2

statisticallyinsignificant.blog/2025/09/02/m...

Adam Kleczkowski @aklecz.bsky.social · Sep 2

And, quoting from Laura Spinney's Pale Rider:

"...trust is not something that can be built up quickly. If it is not in place when a pandemic declares itself, then however good the information being circulated, it probably won’t be heeded."

1

Adam Kleczkowski @aklecz.bsky.social · Sep 2

It is indeed one of the most important graphs of the 2020s.

Sadly, for their own reasons, politicians missold the vaccination and that created a backlash. We lost people's trust.

This is what I wrote in June 2021:

statisticallyinsignificant.blog/2021/06/07/c...

1

Adam Kleczkowski @aklecz.bsky.social · Sep 2

Defining "succeed" is the key here. It is already an excellent microblogging site, with (so far) largely balanced views, little spam and no ads, and - most importantly - a lot of interesting, dedicated people writing interesting posts.

I now hardly ever read X and find Threads/Instagram unusable.

1 1 6

Adam Kleczkowski @aklecz.bsky.social · Sep 2

🤣

New-Cleckit Dominie @ncdominie.bsky.social · Sep 2

There are signs in the lift foyer directing prospective computer science students to Level 11.

So far I've resisted the temptation to get out a biro and append the words "(base ten)".

2 3

Reposted by Adam Kleczkowski

Raywat Deonandan @deonandan.bsky.social · Sep 2

I love looking at this graph, showing it to my students, and sharing it on social media, even though it unerringly brings out the trolls.

This is the efficacy curve of Pfizer's mRNA COVID vaccine.

This, people, is what ended the emergency phase of the COVID pandemic. Despite what RFK Jr says.

31 530 1.5K

Adam Kleczkowski @aklecz.bsky.social · Sep 2

Low sensitivity and specificity are a big problem with AI-powered tools.

The application determines whether such levels can be tolerated. But if it sends the wrong message, better not to implement.

My blog from the COVID era:
statisticallyinsignificant.blog/2020/09/22/f...

Peter Tennant @pwgtennant.bsky.social · Sep 2

A very important caveat to the "AI stethoscope" that was completely ignored by many news outlets.

Yet a high false positive rate can make a screening tool more harmful than useful.

How can we encourage health journalists to be more critical of AI press releases?

Cc @mgtmccartney.bsky.social

Euronews @euronews.com · Sep 1

About two-thirds of patients who were flagged by the AI stethoscope as potentially having heart failure did not actually have it.

2

Reposted by Adam Kleczkowski

Prof Katherine Schofield 🇬🇧🇦🇺🇪🇺 @katherineschof8.bsky.social · Sep 2

These are incredibly useful tips for all teachers and professors starting a new academic year...

Mike Dickison @adzebill.bsky.social · Aug 31

Wikipedia editors trying to fend off the onslaught of AI crap have crowdsourced some telltale signs of LLM-generated writing; it might be handy for editors and proofreaders generally. Thanks to @ellenrykers.com for pointing me to it. en.wikipedia.org/wiki/Wikiped...

Wikipedia:Signs of AI writing - Wikipedia

en.wikipedia.org

5 28 58