Guillaume Cabanac ⟨here and elsewhere⟩ @gcabanac Twitter profile | Pikagi

Pikagi

Guillaume Cabanac ⟨here and elsewhere⟩

@gcabanac

4,691

Followers

1,579

Following

1,499

Media

18,564

Statuses

‘Deception Sleuth’ #Nature10 Scientist • CS Prof. @UT3PaulSabatier @InstUnivFr • Scientific Text Mining •

Toulouse, France

https://t.co/i5oiFuNWk3

Joined January 2011

Don't wanna be here? Send us removal request.

Pinned Tweet

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

The ‘Problematic Paper Screener’: flags unreliable passages in publications, supports a decentralised re-assessment of problematic papers, helps to decontaminate the scientific literature. That's my research as @InstUnivFr member.

Tweet card media

Guillaume Cabanac, « dépollueur scientifique » | Le blob, l'extra-média

Les articles frauduleux « polluent » la science. Guillaume Cabanac veut la dépolluer. À cet effet, cet enseignant-chercheur a conçu un logiciel qui passe au crible les publications scientifiques à la...

3

55

112

Last Seen Profiles

@dairysnake

@HillmanJE

@na9974918633896

@Memon_71

@_Mike_McCabe

@curry_boz

@Si_Gilliam

@Ohiogirl794

@_a_aurora_

@Buddhi99

@SE_AfghanWGH

@hthai9324

@AnkuSat

@strengthshep

@AdolfogusResist

@25Amwlh

@MaRie2424247493

@BlackAndGoldPod

@SnyderJero56180

@kazuki_matsuiwa

@spurge_

@FantaVicki

@C_Chakamian

@KURIERat

@BoonKlinsuwan

@csestanovich

@gay_bator

@SHPAthletics

@HCTrudo

@Akadem_Campus

@bartholinite

@Drumdums

@KSA_krispykreme

@Bamfordgolf

@FinalBossFutbol

@_twio

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

🤖 So #ChatGPT wrote the first sentence of this @ElsevierConnect article. Any other parts of the article too? How come none of the coauthors, Editor-in-Chief, reviewers, typesetters noticed? How can this happen with regular peer-review?

Tweet media one

136

1K

5K

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

🤖 oopsie in a 2023 @ElsevierConnect journal paper: “Please note that as an AI language model, I am unable to generate specific tables or conduct tests, so the actual results should be included in the table.” 🤡 #ChatGPT

Tweet media one

30

378

2K

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 year

🤖 #ChatGPT wrote the ‘Discussion’ section of an @MDPIOpenAccess paper in the 𝘛𝘰𝘹𝘪𝘯𝘴 journal?! The authors may have failed to remove the “Regenerate Response” label of the button when they copy-pasted ChatGPT's output 🤡 #fail Reported on @PubPeer

Tweet media one

59

442

1K

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

Just out @arxiv : “Tortured phrases: A dubious writing style emerging in science. Evidence of critical issues affecting established journals” w/ Labbé & Magazinov Probable systematic manipulation of the publication process of Microprocessors in Microsystems

Tweet media one

Tweet media two

Tweet media three

Tweet media four

34

148

350

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

🤖 #ChatGPT 's “As an AI language model, I...” telling fingerprint in an @ElsevierConnect article of Radiology Case Reports.

Tweet media one

6

69

222

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

@Spottingthespot @ElsevierConnect @RetractionWatch @HollyElse @byamandahoover @gvconroy @JeffreyBrainard @joelving @rhysblakely I post on PubPeer and the most blatant cases end up here on Twitter with the publishers tagged. This is pro bono. Up to them now...

3

1

146

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

@KasperKepp @ElsevierConnect @RetractionWatch @HollyElse @byamandahoover @gvconroy @JeffreyBrainard @joelving @rhysblakely I reported it 9 months ago already: They must be investigating…

3

0

131

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

🤖 #ChatGPT misuse in a @IEEEorg conference article. What else was generated in such papers? Why did peer review fail so badly? What will AI learn from these questionable research articles? Public money well spent? Are you listening @ComputerSociety ?

Tweet media one

8

29

114

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

@DVHenkelWallace @JohnDCook @ElsevierConnect @RetractionWatch @C0PE For potential cases of misconduct like this, I would like the publisher @ElsevierConnect to post the anonymised referee reports on @PubPeer so that the community check how thorough the evaluation process was.

1

7

110

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

#interdisp “Scientists who leave the safe haven of their home discipline to explore the uncharted territory that lies outside and between established disciplines are often punished rather than rewarded for following their scientific curiosity.” @UdiShapiro

Tweet media one

2

51

111

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

Other cases with “As an AI language model, I” or “Regenerate response” that are fingerprints of #ChatGPT 🤖 in peer-reviewed papers, signs of poor evaluation process 😧, and translate into unreliable ‘scientific’ bricks of knowledge

1

15

102

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

🤖 Is #ChatGPT more honest than the authors: “As an AI language model, I cannot determine how good your results are without more context.” 😵‍💫 Published in a 2023 @IEEEorg conference proceedings. This is claimed to be peer-reviewed research 🤦‍♂️ More context:

Tweet media one

6

37

97

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

@ScienceParle_SW @ElsevierConnect @RetractionWatch @HollyElse @byamandahoover @gvconroy @JeffreyBrainard @joelving @rhysblakely Here is how

0

6

96

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

4 months

Retracted hoax 🤰🤚🌈👦🇸🇬: “Left-Handed Mothers and LGBTQ+ Child Acceptance in Singapore: Exploring the Link through Early Life Rejection” — A sting operation to question the peer review process of the Journal of Education, Society and Behavioural Science

Tweet media one

7

37

89

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

Value of post-publication peer review, Exhibit 326: “The authors have perhaps good skills in using #ChatGPT , but could be a public menace if by some misfortune they are hired by an hospital 🏥.” in a @SpringerNature conference article. HT Sylvain Bernès

Tweet media one

Tweet media two

4

18

88

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

6 months

😵 Published in October 2023 with **29 retracted references** in its biography: a @FrontOncology review on MicroRNAs, long non-coding RNAs, and circular RNAs and gynecological cancers. How reliable is this review article? #FeetOfClay #DominoEffect

Tweet media one

8

30

86

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

27 days

🖼️ #GenerativeAI will soon flood the scientific literature with nonsensical-yet-colourful images 🍭: “figure seem[s] nice and clear” says Reviewer 3 😵 of this @MDPIOpenAccess Pharmaceutics article, see

Tweet media one

Tweet media two

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

@AlexandraDelbot @ONeyrolles @internetarchive @altmetric @RetractionWatch @joelving @HollyElse @byamandahoover @JeffreyBrainard @n6g6c @gvconroy @DalmeetS @cdurandparenti @rhysblakely 🥴 Apparently the ‘Chimeric Rat’ Review Article passed peer review (as implements by Frontiers in Cell and Developmental Biology) with at least 2 reviewers, see the comments on the right hand side... (PDF: )

Tweet media one

3

7

24

3

28

85

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

“First they ignore you, then laugh at you, then they fight you then you win.” – Ghandi Science wins, thank you for your dedication and hard work Dr. Elisabeth Bik @MicrobiomDigest !

4

17

83

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 month

🤖 #ChatGPT “regenerate response” fingerprint in reviewers' reports: I found some in @MDPIOpenAccess journals, e.g., and . It seems that fancy adjectives are good predictors, too... Is peer review doomed? 🤯

Tweet media one

Tweet media two

@MishaTeplitskiy

Science of Science

@MishaTeplitskiy

1 month

Lots of people in CS are (almost surely) GPT-ing their peer reviews

Tweet media one

53

1K

6K

6

18

80

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

11 months

#ChatGPT wrote passages of this @IEEEorg conference paper. The caveat “As an AI language model...” is characteristic 😵‍💫. @ComputerSociety should audit the peer review process held, reassess the entire proceedings, and retract unreliable papers.

Tweet media one

7

32

80

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

🤩 My application to the @InstUnivFr Fundamental Chair programme is selected! I'll extend my #interdisciplinary and #collaborative work on #researchIntegrity as a member of the ‘Institut universitaire de France’ for 5 years.

Tweet card media

Institut Universitaire de France - Wikipedia

en.wikipedia.org

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

The ‘Problematic Paper Screener’ flags 7,658 articles cited 48k times as problematic with 7 detectors 😵. Your human re-assessment welcome @PubPeer for 4,551 papers with tortured phrases. Let's depollute the scientific literature 🔎

Tweet media one

Tweet media two

Tweet media three

Tweet media four

0

8

18

15

7

76

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

En 2002, je suppliais ma prof d'algo @IutToulouse3 de m'accueillir en stage dans son labo de recherche. Atypique à bac+2. Une immersion à @IRIToulouse qui a déterminé mon projet professionnel. CJ, ton « oui, je te prends » a changé ma vie. #donnezunechance

Tweet card media

Guillaume Cabanac, traqueur de fake science

"Corrosif nucléique" pour "acide nucléique". "Conscience contrefaite" à la place de "intelligence artificielle"... Ce sont les expressions "torturées" que Guillaume Cabanac traque dans les

lejournal.cnrs.fr

3

17

74

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

6 months

🤖 Authors put #ChatGPT 's “Regenerate Response” in an @ElsevierConnect article of the Natural Language Processing Journal. I reported the issue in July . Elsevier operated a *stealth revision* in September. Who cares about a proper corrigendum? #VOR #trust

Tweet media one

Tweet media two

1

24

70

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

8 years

Copyright madness #rule40fail #Rio2016

Tweet media one

12

103

68

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

“Digital humanities—A discipline in its own right? An analysis of the role and position of digital humanities in the academic landscape” just out in @JASIST #DH

Tweet media one

0

22

67

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

📚 “Humour is the interdisciplinary denominator in science” (Kohn, 1982): Start the new year with my collection of 130 papers featuring humour, whimsy, easter eggs, and other bizarreries 🤔🤓🤪😵‍💫 “Unconventional academic writing” #humour #science #research

Tweet media one

Tweet media two

Tweet media three

Tweet media four

2

18

67

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

🤔 A surprisingly frequent typo: ‘prosciutto’ (an Italian 🇮🇹 word for ‘dried ham’ 🐖) should read ‘perovskite’ 🪨 . Why is this happening? Strange that nobody noticed: coauthors, EiC, reviewers, typesetter, readers. Posts on @PubPeer

Tweet media one

Tweet media two

Tweet media three

11

16

66

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

I am now a proud patron of Elisabeth Bik (Harbers-Bik LLC @MicrobiomDigest ), Science Integrity Detective on @patreon . “Efficiency is doing things right, effectiveness is doing the right thing.” — Peter Drucker. She does both.

1

8

58

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

Just out: “Prevalence of nonsensical algorithmically generated papers in the scientific literature” w/ Cyril Labbé in @JASIST . We report cases of SCIgen-generated publications. 😵 Companion website + thread:

Tweet media one

Tweet media two

Tweet media three

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

@DalmeetS @NatureIndex @jrenauardevol @IEEEorg @tianyin_xu @vj_chidambaram @heyelbs There's a long thread of misconduct cases, mostly in computer science, () that we posted on @PubPeer (). No reaction from the publishers involved to date AFAIK. Care for sunlight as disinfectant, anyone?

0

7

9

7

29

58

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 year

The Problematic Paper Screener turns 2 today 🎂. Nine detectors screen the scientific literature using @DSDimensions . The PPS flagged 11k problematic papers using ‘fingerprints’ contributed by dozens of @PubPeer sleuths. To be reassessed: 4k papers… 🕵️

Tweet media one

Tweet media two

1

24

58

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

The ‘Feet of Clay’ Detector detects papers citing unreliable literature 😨: retracted and problematic papers commented on @PubPeer . So far: 3k Feet of Clay papers found, with alarming concentrations in some venues. Part of the Problematic Paper Screener

Tweet media one

Tweet media two

Tweet media three

Tweet media four

3

24

57

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

👀 Read my colleague and friend Prof. Chérifa @BoukacemZeg ’s World View in @Nature : “ #Predatory journals entrap unsuspecting scientists. Here’s how universities can support researchers” HT @UnivLyon1 #metascience #scienceofscience

Tweet card media

Predatory journals entrap unsuspecting scientists. Here’s how universities can support researchers

Nature - Training from institutions on publishing norms could help to thwart predatory publishers.

2

21

57

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

5 months

🇫🇷 « Signalement d'articles scientifiques non fiables par traque de marqueurs de Fake Science », exposé invité à la Journée science ouverte : logiciels libres et fouille de textes, organisée hier par @SciOuverte_CNRS à la délégation @CNRS Paris Michel-Ange

Tweet media one

2

19

51

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 year

@mixlamalice Suis resté à l’ère Giscard dans mon bureau d’IUT créé en 1968. Ce qui paraît plus récent a été en partie récupéré dans des bennes sur la fac (certains services admin virent des meubles des années 2000 : du rebut pour eux, un progrès significatif en confort pour nous).

Tweet media one

6

2

50

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

🕵️🤖 “Academic sleuth sniffs out #ChatGPT fakery in research papers” in @TheTimes today, interviewed w/ @mattjhodgkinson @MicrobiomDigest @ElsevierConnect about this @PubPeer report I wrote last August. Read @rhysblakely 's article:

4

11

50

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

4 months

My two cents in a @TheTimes 🇬🇧 paper titled “Record year for bad science with 10,000 studies retracted” by @rhysblakely . Also featuring @MicrobiomDigest @Nature and @Hindawi @WileyGlobal @HarvardHBS .

Tweet media one

4

28

49

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

The ‘Problematic Paper Screener’ has just turned one 🎂. It flags 5,716 articles as problematic with 7 detectors. They are cited 38k times 😒 Only 1,516 reported on @PubPeer . Let's depollute the scientific literature: 3,226 tortured papers to re-assess 🔎

Tweet media one

Tweet media two

Tweet media three

Tweet media four

3

20

49

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

🤖 #ChatGPT 's “Regenerate Response” in a @SpringerNature article of ‘Environmental Science and Pollution Research’ What else did it generate here? Sad for science but not surprised. See the journal-level report I posted in June 2023

Tweet media one

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

10 months

🏚️ 66+ problematic papers flagged in the “Environmental Science and Pollution Research” journal published by @SpringerNature . They contain tortured phrases, questioned references, and other anomalies 😵‍💫. Publisher + @C0PE informed. Retract/correct ASAP.

Tweet media one

Tweet media two

0

12

22

3

24

48

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

A sad day for the #scientometrics community: Prof. Tibor Braun passed at the age of 90. He founded Scientometrics in 1978 ; he invited me to join its board. I enjoyed his correspondence, so inspirational. We happen to be born on March 8th, 50 years apart 😢

Tweet media one

2

18

48

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

📰 Our ‘sneaked in references’ preprint featured in @RetractionWatch : “How thousands of invisible citations sneak into papers and make for fake metrics” by @DalmeetS w/ @GinnyLDN @CrossrefOrg and @altmetric @DSDimensions mentioned.

@lonnibesancon

Lonni Besançon 🇫🇷 🇸🇪

7 months

Latest preprint out!🔔 With @gcabanac C. Labbé, and A. Magazinov, we document a new way of manipulating citation counts through Crossref metadata registration. References can be ‘sneaked in’ while publishers register papers and their bibliographies. 1/n

Tweet media one

9

57

100

1

35

47

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

Mixing oil with vinegar🤯. Sold as a peer-reviewed publication in a @SpringerNature journal with an Impact Factor 😵‍💫. See Cudos to all @PubPeer sleuths. #AbusedScience #OpenCallForInvestigation

Tweet media one

Tweet media two

Tweet media three

5

21

46

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

11 months

😵‍💫 “thin movies” instead of “thin films” / “X-beam diffraction” instead of “X-ray diffraction” / “surface unpleasantness” instead of “surface roughness”... #TorturedPhrases in @ElsevierConnect Optik journal CC @ElsevierPhysics . See @PubPeer my report

Tweet media one

3

8

47

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

🤖 #ChatGPT 's “Regenerate Response” in the bibliography of a @PLOSONE article. Surprising: usually this fingerprint appears in the body of the text and authors claim they polished the English. Not much to polish in a reference section. Hallucination?

Tweet media one

Tweet media two

4

12

47

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

Pour ma part, l’informatique me permet de traquer des erreurs et fraudes dans la literature scientifique. Mon programme analyse 120 millions d’articles scientifiques chaque nuit depuis le serveur d’ @IRIToulouse . 3 articles non fiables par jour en moyenne.

Tweet card media

Guillaume Cabanac, « dépollueur scientifique » | Le blob, l'extra-média

Les articles frauduleux « polluent » la science. Guillaume Cabanac veut la dépolluer. À cet effet, cet enseignant-chercheur a conçu un logiciel qui passe au crible les publications scientifiques à la...

@dptinfo_fsi_ut3

Département informatique FSI de l’UT3

@dptinfo_fsi_ut3

2 years

Opter pour la spécialité NSI, c'est choisir le système algorithmique, les langages de programmation et la gestion des bases de données. Avec de larges débouchés, comme l’informatique, la santé, la littérature ou encore le développement durable.

2

6

7

6

14

46

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

4 months

🏅 to @IOPPublishing : @EggletonKim and Team. They monitor @PubPeer , retract problematic papers swiftly, issue retraction notices acknowledging PubPeer and linking to post-publication reports … Now transferring the APCs of retracted papers to charities! 👏

@RetractionWatch

Retraction Watch

@RetractionWatch

4 months

Publisher donating author fees from retracted articles to charity

Tweet media one

0

7

16

2

11

45

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 years

“Sci-Hub provides access to nearly all scholarly literature” by @dhimmel arielsvn @gedankenstuecke @GreeneScientist

Tweet card media

Sci-Hub provides access to nearly all scholarly literature

greenelab.github.io

1

37

44

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 years

I deleted my @Academia and @ResearchGate accounts today 🤑. Read @SarahEBond 's @Forbes article to find out why 👏 #OA

Tweet media one

8

39

43

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 years

Je soutiendrai mon HDR intitulée « Interroger le texte scientifique » à @IRIT_UMR5505 le 8/12 à 10h #habilitation àdirigerdesrecherches

Tweet media one

Tweet media two

9

11

45

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

Details here:

0

1

44

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

6 months

🧑‍⚖️ Retracted at last! I reported this #SCIgen nonsense in July 2022, @SpringerNature retracted in September 2023. Were the 14 months of publisher investigation even required? I let you make your own mind… HT @SmutClyde for letting me know; no news from SN

Tweet media one

Tweet media two

Tweet media three

Tweet media four

2

11

42

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 years

I am a proud father of 3 now: 👧 Lise is nearly 6 y.o., 👧🏻 Agathe is nearly 3 y.o., and my son Jean 👶🏻 was born today at 2:17 a.m. 🎉 #happy 💑

22

1

43

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

We posted on @PubPeer 7 cases of plagiarism, image theft, and use of tortured phrases (e.g., ‘credulous/innocent Bayes’ in lieu of ‘naïve Bayes’; ‘counterfeit consciousness’ vs. ‘AI’) in Microprocessors in Microsystems. Problematic papers tabulated here

Tweet media one

Tweet media two

Tweet media three

Tweet media four

1

9

42

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

4 months

Paper on #fraud detection with copy-paste ‘Regenerate Response’ button of #ChatGPT UI. Priceless publication of the @IEEEorg @IEEEXplore @ComputerSociety portfolio 🤦 Re IEEE, see and 123+ proceedings 😵‍💫

Tweet media one

6

18

42

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 year

New preprint: “The Geography of Retracted Papers: Showcasing a @CrossrefOrg – @DSDimensions –NETSCITY [] Pipeline for the Spatial Analysis of Bibliographic Data” w/ @alexcla99 @ljegou @GeoMaisonobe 🗺️ #retractions #inprogress #sti2023

Tweet media one

2

25

42

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

New scientific integrity case in France: @Ifremer_fr reports 10 papers for “forged authorship, usurped or theft of authors' names and affiliations, and unethical practice of journals” here Bravo for posting under an institutional @PubPeer account.

Tweet media one

2

14

41

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

On @ARTEfr in French: a 3-minute summary of my @InstUnivFr @IRIToulouse research on the decontamination of science. From flagging ‘tortured phrases’ 😵‍💫 to spotting #ChatGPT fingerprints 🤖 in peer-reviewed articles. ➡️ Problematic Paper Screener 🕵️‍♂️

@28minutes

28 minutes

7 months

Des auteurs “pris la main dans le chat” ! Si on blâme souvent les étudiants pour l’utilisation de l’intelligence artificielle, les chercheurs et écrivains ne sont pas en reste… Alix Van Pée nous en parle. 🤖

4

10

32

3

18

40

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 month

😵‍💫 ‘Outrageous Gradient Boosting’ instead of XGBoosting: new tortured phrase recorded. Also included ‘mind tumor’ 🤯. Found in the book ‘Predictive Modeling in Biomedical 🏥 Data Mining and Analysis’ sold $126 by @ElsevierConnect @ElsevierNews 🤑

Tweet media one

Tweet media two

3

6

40

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

Retracted: 2 papers by Max Planck (the 1918 Nobel Prize Winner in Physics 👨‍🎓) marked ‘withdrawn due to copyright/article violation’ in @Springer_Shop . Can @SpringerNature and @maxplanckpress be more specific please? What happened? 🤔

Tweet media one

Tweet media two

2

13

40

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

The ‘Problematic Paper Screener’ flags 6,308 articles cited 43k times as problematic with 7 detectors 😵. Your human re-assessment welcome @PubPeer for 3,629 papers with tortured phrases. Let's depollute the scientific literature 🔎

Tweet media one

Tweet media two

Tweet media three

Tweet media four

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

The ‘Problematic Paper Screener’ has just turned one 🎂. It flags 5,716 articles as problematic with 7 detectors. They are cited 38k times 😒 Only 1,516 reported on @PubPeer . Let's depollute the scientific literature: 3,226 tortured papers to re-assess 🔎

Tweet media one

Tweet media two

Tweet media three

Tweet media four

3

20

49

4

6

38

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

🤖 #ChatGPT 's “Regenerate response” in a PhD dissertation defended in 2023 at the Stevens Institute of Technology @FollowStevens based in Hoboken, NJ, USA 🇺🇸 . Alas, this telling fingerprint also appeared in peer-reviewed articles:

Tweet media one

2

9

39

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

I’m pleased and honoured to work with Dorothy Bishop @deevybee 🕵️‍♀️, Emeritus Professor at @UniofOxford @OxExpPsy , on the decontamination of the scientific literature. A challenging battle against #papermills …

@lemondefr

Le Monde

9 months

Dorothy Bishop, une détective spécialiste de la mauvaise science

4

11

25

2

7

38

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

8 months

📰 @WIRED covered that case of a #ChatGPT fingerprint I found in an @ElsevierConnect paper. See “Use of AI Is Seeping Into Academic Journals—and It’s Proving Difficult to Detect” @byamandahoover feat. @MicrobiomDigest , @ResnikDavidB , Heather Desaire

Tweet card media

Use of AI Is Seeping Into Academic Journals—and It’s Proving Difficult to Detect

Ethics watchdogs are looking out for potentially undisclosed use of generative AI in scientific writing. But there’s no foolproof way to catch it all yet.

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

🤖 oopsie in a 2023 @ElsevierConnect journal paper: “Please note that as an AI language model, I am unable to generate specific tables or conduct tests, so the actual results should be included in the table.” 🤡 #ChatGPT

Tweet media one

30

378

2K

1

9

38

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

Just out: “Day-to-day discovery of preprint–publication links” in 𝘚𝘤𝘪𝘦𝘯𝘵𝘰𝘮𝘦𝘵𝘳𝘪𝘤𝘴 w/ @dora_oikonomidi @BOUTRON1 , an interdisciplinary research @IRIToulouse and @CRESS_U1153 . Our linker informs @CovidNMA daily on #COVID19 🦠 at @CochraneFrance .

Tweet media one

Tweet media two

Tweet media three

Tweet media four

4

21

38

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

The Problematic Papers Leaderboard 🎃: 615 articles from flagship publishers tabulated in the 𝘗𝘳𝘰𝘣𝘭𝘦𝘮𝘢𝘵𝘪𝘤 𝘗𝘢𝘱𝘦𝘳 𝘚𝘤𝘳𝘦𝘦𝘯𝘦𝘳 and associated to @PubPeer comments. Publishers must react and investigate, especially @C0PE members CC @ALPSP

Tweet media one

Tweet media two

Tweet media three

Tweet media four

3

19

37

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

5 months

🧑‍🏫 “Fake Science: Misconduct Galore and Proposed Counterattack”: my invited talk at @LaasCNRS @CNRS_Toulouse @Univ_Toulouse on research misconduct and actionable preventive measures HT funding from @InstUnivFr @_Nano_Bubbles @ERC_Research

Fake Science: Misconduct Galore and Proposed Counterattack

The issue of ‘fake’ does not spare science. This conference will shape the landscape of misconduct that contaminate the scientific literature. It will also stress counterattacks to prevent these by...

2

15

35

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

📨 Invitation from @SciReports by @SpringerNature : “Scientific Reports is committed to providing a rapid and fair review process […] receive your report within 10 days.” 🏃🥵 I declined and suggested the Editor to attend to these post-publication reports

Tweet media one

1

12

36

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 months

“Universities must declare all their retractions and launch investigations into misconduct cases; a @Nature analysis reveals that since 2021 there have been more than 17,000 retractions with Chinese co-authors 🇨🇳.” by @SmritiMallapaty

Tweet card media

China conducts first nationwide review of retractions and research misconduct

Nature - Universities must declare all their retractions and launch investigations into misconduct cases; a Nature analysis reveals that since 2021 there have been more than 17,000 retractions with...

1

18

36

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

🎥 « Pollution de la littérature scientifique : détection participative d'expressions torturées révélatrices d'articles frauduleux », ma conférence à @UT3PaulSabatier du 14 juin est en ligne. #plagiat #papermills #int égrité #science

Tweet card media

cMED: "Pollution de la littérature scientifique:détection....d'arti...

Par Guillaume Cabanac:Guillaume Cabanac est maître de conférences habilité à diriger des recherches (HDR) en informatique à l'université Toulouse III – Paul ...

www.youtube.com

@UT3PaulSabatier

UT3 Paul Sabatier

@UT3PaulSabatier

2 years

En ce moment conférence médiation éthique et déontologie animée par @gcabanac sur la thématique de la pollution littéraire scientifique et la détection d'articles frauduleux. #papermills #problematicpaperscreener

Tweet media one

0

4

13

0

12

35

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 years

”The intellectual structure of Information Science 2006–2015“ in (Yang et al., JoI 2016, p. 143) #infsci #bir2017

Tweet media one

2

24

33

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 year

🤖 #ChatGPT : “As an AI language model, I...” found in preprints and papers! A fingerprint for the ‘suspect’ detector of the Problematic Paper Screener 🕵️, thanks for the tip @ClearSkiesAdam HT @mattjhodgkinson

Tweet media one

Tweet media two

@ClearSkiesAdam

Adam Day

@ClearSkiesAdam

1 year

Tortured phrases for twitter bots? @gcabanac

1

1

8

2

11

33

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

A #SCIgen computer-generated paper appeared in 𝘗𝘳𝘰𝘤𝘦𝘥𝘪𝘢 𝘊𝘰𝘮𝘱𝘶𝘵𝘦𝘳 𝘚𝘤𝘪𝘦𝘯𝘤𝘦 ( @ElsevierConnect ) in February 2021.

Tweet media one

Tweet media two

Tweet media three

Tweet media four

2

14

33

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

5 years

Call for Journal Papers: Special Issue on “Bibliometrics and Information Retrieval” to appear in the Scientometrics journal w/ @gcabanac @Philipp_Mayr @iFromm as Guest Editors. Deadline 📝: end of September 2019 #scientometrics #IR

Tweet media one

5

29

33

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

« Problematic Paper Screener : détection d'expressions torturées révélatrices d'articles frauduleux » interview diffusée dans le 19/20 de @F3Occitanie du 07/02/2022. Une recherche @UT3PaulSabatier @IRIToulouse . Compléments : et

3

18

33

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

🙉 Stunning silence from @SpringerNature re my report of 33+ problematic papers in @SciReports . So, to the new “invitation to review [...] within 10 days” 📧 I received today, I replied once again: correct/retract these first!

Tweet media one

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

📨 Invitation from @SciReports by @SpringerNature : “Scientific Reports is committed to providing a rapid and fair review process […] receive your report within 10 days.” 🏃🥵 I declined and suggested the Editor to attend to these post-publication reports

Tweet media one

1

12

36

0

15

33

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

6 months

🏆 #HiCi Clarivate's Highly Cited Researchers list of 2023 is out today . They used ‘tortured phrases’ as a bad signal when evaluating the nominees, as shown in box 3 😵‍💫Crazy indeed: +

Tweet media one

2

11

33

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

11 months

⚠️ 56+ problematic papers flagged in _Multimedia Tools and Applications_ 📽️🎙️📺 published by @SpringerNature . They contain tortured phrases, questioned references, and other anomalies. Retractions/corrections needed. Journal-level report by the PPS:

Tweet media one

5

19

32

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

5 months

🗞️ Our latest preprint covered in Le Monde @lemondefr 🇫🇷 with this title: ‘Research: a traffic of fake bibliographic references uncovered’. Fulltext available upon request.

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

📰 Our ‘sneaked in references’ preprint featured in @RetractionWatch : “How thousands of invisible citations sneak into papers and make for fake metrics” by @DalmeetS w/ @GinnyLDN @CrossrefOrg and @altmetric @DSDimensions mentioned.

1

35

47

0

18

32

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

10 months

47+ problematic papers flagged in @PLOSONE by the Problematic Paper Screener . They contain tortured phrases, questioned references, and other anomalies 😵‍💫. Retractions/corrections needed++ @PLOS .

Tweet media one

Tweet media two

3

11

32

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

3 years

“Refcat: The @internetarchive Scholar Citation Graph” preprint by Martin Czygan, @helgeho , Bryan Newbold

Tweet media one

0

13

32

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

19 days

🙏 Thanks to the anonymous readers who suggest new #TorturedPhrases to monitor: ‘attractive stirrer’ ⚗️ and ‘the taking after’. Both published in @SciReports , part of the @Nature portfolio. See for yourself: and

Tweet media one

Tweet media two

7

10

31

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

👩‍🏫 “Errors and Misconduct in Biomedical Research / Correction of the Scientific Literature” by Dr. Elisabeth Bik @MicrobiomDigest (see her blog ) at the #Inserm277 workshop on meta-research: . What a treat! HT @Inserm

Tweet media one

Tweet media two

Tweet media three

Tweet media four

1

9

31

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

5 years

Science 📚 + Cities 🌆: “Peripheral forces: The growing impact of second-tier cities is narrowing the gap in research production” just out in @NatureIndex w/ @GeoMaisonobe @ljegou #GeographyOfScience #scientometrics

Tweet media one

1

17

31

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

9 months

🤬 Chapters of @MITpress #OA book sold $15 each at @IEEEXplore @ComputerSociety despite the @CreativeCommons BY-NC-ND license! Paging the editors Prof. Biagioli @UCLA_Law & Dr. @xandao_sound , OA funder @UCDavisLibrary , authors @IvanOransky , @BorisBarbour …

Tweet media one

Tweet media two

Tweet media three

4

8

30

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

4 years

How it started: How it’s going:

Tweet media one

Tweet media two

2

0

30

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 years

“Order of #authorship was determined by proximity to #tenure decisions.” Wise and sensitive academics.

Tweet media one

0

5

30

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 month

🤖 #ChatGPT caveat “As of my knowledge cutoff in September 2021, ...” spotted by @mmmpsych and I, see the @PubPeer reports and other telling LLM fingerprints found in peer-reviewed articles . Reviewers asleep?

@SmutClyde

Smut Clyde, X-Ray Haruspex

1 month

ChatGPT, you are the expert on Artificial Stupidity, write a scholarly manuscript on that topic for me to sign. Include a References section. They don't have to be real.

Tweet media one

Tweet media two

0

5

28

4

8

29

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

On top of buying authorship (see @author_for_sale ), buying citations (see ) scammers can afford a @Wikipedia page for $1500-$2500 to impress grant agencies and promotion committees 🤑 #spam #gamingTheMetrics #nolimit

Tweet media one

1

19

29

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

16 days

🧑‍🎤 3 tortured phrases for “heavy metal” tabulated in the Problematic Paper Screener: . See . Bonus: a remarkable thread feat. @raphavisses , @SmutClyde , and Sylvain Bernès CC @IUPAC 🎸

Tweet media one

@olivier_pourret

[email protected]

@olivier_pourret

16 days

Evolution of the number of publications using the term “heavy metal*” in their title, comparing @OpenAlex_org @DSDimensions @Scopus @webofscience will soon update my paper

Tweet media one

5

5

11

4

10

29

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

2 months

@Cacciamani_MD @ElsevierConnect @RetractionWatch @byamandahoover @HollyElse @joelving @JeffreyBrainard @rhysblakely @DalmeetS I've been monitoring #ChatGPT fingerprints that shouldn't have survived peer review 😲. The problematic papers published in non-predatory (?) journals make a corpus of 82 @PubPeer posts here:

0

4

29

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 month

Fellow academic @cgrodriguezr had some doubts about a an article and asked for my advice. My recommendation: install the @PubPeer browser extension and discover that some references got questioned, e.g.

Tweet media one

@cgrodriguezr

Cristian Rodriguez

1 month

Perhaps @gcabanac can help me figuring this one out

1

0

1

2

12

29

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

5 years

The #issi2019 Doctoral Forum is starting in Rome 🇮🇹 w/ @csugimoto @AlesiaZuccala @lariviev et al. Feedback appears realtime here . Find the full conference programme here

Tweet media one

Tweet media two

2

9

29

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

6 years

Excited to join the Editorial Board of @JASIST , the Journal of the Association for Information Science & Technology

Tweet media one

Tweet media two

Tweet media three

5

5

29

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

11 months

🤖 “Regenerate response” in the introduction of a 2023 paper in the European Journal of Mass Spectrometry, a @SAGEPublishers journal... Neither the coauthors, editor, reviewers, typesetters noticed the caption of this #ChatGPT button, copied by mistake...

Tweet media one

0

13

29

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

8 months

😲 Question to the authors: does the web interface of @MDPIOpenAccess reflect the report you received from Reviewer 1 in @ApplSci ? Credit to @GillesHub for spotting this.

Tweet media one

2

11

28

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

5 years

I am honoured to join the @dblp_org Advisory Board and on my way to the 2-day installation meeting in legendary @dagstuhl . Help me to relay the #CS community's feedback on @dblp_org : Please voice your suggestions, concerns, praises ... in the thread ⬇️

3

8

28

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

7 months

🙅‍♂️ Just declined: invitation to review for @Water_MDPI . 580 problematic MDPI papers listed in the ‘Problematic Paper Screener’ and I just reported issues in @Water_MDPI : article with 5 tortured phrases

Tweet media one

Tweet media two

Tweet media three

2

8

28

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

1 year

😵‍💫 “mind waves” and “cerebrum PC interface”: #TorturedPhrases in a @SpringerNature review article published in 2021, sold 39,95 € and cited 80 times already. When the text reflects blatant plagiarism, what to think of the science inside? PubPeered here:

Tweet media one

Tweet media two

4

13

28

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

4 months

😵 #TorturedPhrases in Information Processing & Management, a @ElsevierConnect journal. 😟 I'm so disappointed: I've been reading IP&M since my MSc years. I reported 5 papers to the EiC, including this: . Also declined the review invitation I had received.

Tweet media one

Tweet media two

3

7

28

@gcabanac

Guillaume Cabanac ⟨here and elsewhere⟩

11 months

🚑 23+ problematic papers flagged in the @HeliyonJournal published by @CellCellPress @ElsevierConnect . They contain tortured phrases, questioned references, and other anomalies 😵‍💫. Retract/correct + transfer APC of $2,100 per unreliable paper to charities

Tweet media one

1

11

27