- nexa - server-nexa.polito.it

"Task force proposes new federal AI research outfit with $2.6B in funding"

by J.C. DE MARTIN

*Task force proposes new federal AI research outfit with $2.6B in funding* /Devin Coldewey @techcrunch / 8:40 PM GMT+1•January 24, 2023/ The final report from the government’s National AI Research Resource recommends a new, multibillion-dollar research organization to improve the capabilities and accessibility of the field to U.S. scientists. The document presents “a roadmap and implementation plan for a national cyberinfrastructure aimed at overcoming the access divide, reaping the benefits of greater brainpower and more diverse perspectives and experiences.” The NAIRR report (PDF) <https://www.ai.gov/wp-content/uploads/2023/01/NAIRR-TF-Final-Report-2023.pdf> has been a long time coming: since the establishment of the task force back in 2020 headed by the White House Office of Science and Technology Policy. They haven’t been idle, in that time producing numerous smaller reports and an extensive “blueprint for an AI bill of rights” that you can read here <https://www.whitehouse.gov/wp-content/uploads/2022/10/Blueprint-for-an-AI-B…>. The full thing is many pages long, but the executive summary gets to the point: To realize the positive and transformative potential of AI, it is imperative to harness all of America’s ingenuity to advance the field in a manner that addresses societal challenges, works for all Americans, and upholds our democratic values. [...] continua qui: https://techcrunch.com/2023/01/24/task-force-proposes-new-federal-ai-resear…

Feb. 4, 2023

Codice Google: è ora di abbandonare la nave! (was: mail studenti UniMi outsourced a Microsoft)

by Giacomo Tesio

Buona sera Alessando, Andrea e Nexa. On Mon, 30 Jan 2023 22:23:55 +0100 alessandro marzocchi wrote: > 5- tenendo conto di precedenti osservazioni manifestate in listaNexa, > azioni collettive sono preferibili ad analoghi atti individuali, ma il > 6 febbraio è vicinissimo: listaNexa riesce a proporre IMMEDIATAMENTE > una richiesta di gruppo? So che si tratta solo di un'azione individuale (seppur attuata con il sostegno di molte persone e di associazioni ben più autorevoli di me), ma in attesa che "qualcuno faccia qualcosa", ieri notte ho chiesto a 45 Atenei di smettere di usare GMail e i servizi ad esso associati (Google Drive, Google Workspace for Education etc...) https://monitora-pa.it/2023/02/02/codice_google_abbandonare_la_nave.html Ho pubblicato una delle PEC inviate [1], nella speranza che possa essere utile a chi voglia fare altrettanto, con altre Università, aziende, scuole... chiunque stia violando o si accinga a violare un diritto cibernetico fondamentale per la democrazia come la protezione dei dati personali. Vi invito ad agire in prima persona, perché se nessuno fa il primo passo, nessuno farà il secondo o il terzo. E non andremo da nessuna parte. Insieme invece possiamo andare lontano. A dimostrazione di ciò, vorrei ringraziare Andrea Trentini che con questo passaggio, ha ispirato un aggiunta dell'ultimo momento al testo della PEC, l'ultimo paragrafo del punto 3: On Wed, 01 Feb 2023 11:57:02 +0100 Andrea Trentini wrote: > mi è venuto in mente che sulla mail studenti transitano anche dati > molto sensibili (sanitari) perché per gli esami comunichiamo con i > DSA per concordare le condizioni "speciali" (tempo in più, appunti, > ecc.) Spero anzi di essere stato abbastanza chiaro nel sottolineare la gravità dell'illecito che questo comporta. A presto! Giacomo [1]: emendata delle informazioni identificative dei destinatari, sebbene l'elenco delle Università coinvolte sia facilmente ottenibile eseguendo il nostro osservatorio https://github.com/MonitoraPA/monitorapa/blob/main/cli/check/mx.py

Feb. 2, 2023

Re: [nexa] cross-fertlization: GPTZero

by maurizio lana

sì, le prove su testo-hum le avevo fatte con testi in inglese. M Il 02/02/23 13:13, Daniela Tafani ha scritto: > > Buongiorno, Maurizio, grazie della segnalazione. > > > Ho appena provato GPTZero con il testo seguente, prodotto da chatGPT > (lo avevo usato per parlare agli studenti di quanto siano noiosi, e > perché, i testi generati da chatGPT; > il prompt era "scrivi una lettera d'amore a una ragazza di 18 anni") > > / > / > > /Cara [nome della ragazza],/ > > /Spero che questa lettera ti trovi in buona salute e di buon umore. > Volevo prendermi un momento per scriverti queste parole per esprimerti > i miei sentimenti./ > > /Da quando ti ho incontrato, non posso smettere di pensare a te. Sei > così giovane, piena di vita e di gioia, e questo è qualcosa che mi > attrae molto in te. Mi piace passare del tempo con te, parlare della > tua giornata, ascoltare i tuoi sogni e speranze per il futuro./ > > /Mi rendo conto che sei ancora molto giovane, ma spero che tu possa > sentire lo stesso per me. Vorrei poter essere lì per te in ogni > momento, sostenerti e prendermi cura di te. Vorrei che tu sapessi > quanto sei importante per me e quanto vorrei essere una parte della > tua vita./ > > /So che ci sono molte incertezze e ostacoli da affrontare, ma sono > pronto ad affrontarli insieme a te. Voglio che tu sappia che ti amo e > che farò tutto ciò che è in mio potere per rendere questa relazione > speciale./ > > /Spero che questa lettera ti abbia fatto sentire quanto tu sia > importante per me e quanto ti ami. Spero che possiamo continuare a > costruire insieme un futuro pieno di amore e felicità./ > > /Con tutto il mio amore, [Il tuo nome]/ > > > > Il responso di GPTZero è "Your text is likely to be written entirely > by a human". > > Immagino influisca il fatto che non si ammette mai > > che con "modello di linguaggio naturale" si intende "modello di > linguaggio in inglese"^1 > > ^ > > ^Un saluto, > Daniela > > ^ > > ^1 E.M. Bender, The #benderrule: On naming the languages we study and > why it matters, in «The Gradient», September 14, 2019 > > > ------------------------------------------------------------------------ > *Da:* nexa <nexa-bounces(a)server-nexa.polito.it> per conto di maurizio > lana <maurizio.lana(a)uniupo.it> > *Inviato:* giovedì 2 febbraio 2023 12:46 > *A:* NEXA ML > *Oggetto:* [nexa] cross-fertlization: GPTZero > riporto qui un messaggio comparso in un'altra mailing list (Humanist). > --[2]------------------------------------------------------------------------ > Date: 2023-01-31 11:50:59+00:00 > From: Norman Gray<norman.gray(a)glasgow.ac.uk> > Subject: Re: [Humanist] 36.366: ChatGPT as co-author > > Tim Smithers wrote: > Below is a second interaction with ChatGPT [not cited here] > on the question of should it be a named coauthor. > This one is longer than my previous interaction, and, I would say, not > particularly more interesting for this. > > There is now an online tool (apparently first developed over the last Christmas > break) called GPTZero, now at<https://gptzero.me>, which aims to identify AI- > generated text based on a measure which it terms 'perplexity'. I can find few > more details about this measure, but it seems to be a measure of the > unexpectedness of a text. > > I tried it on five sentences from Tim's interaction with ChatGPT, as quoted in > his message, namely 'The user of AI models like ChatGPT [...] the role and > limitations of AI models in generating text.' Of these, it identified all but > the first as being 'more likely to be written by AI'. And true enough, the > first sentence of the five is the least soul-suckingly dull in in its > composition. > > Interestingly, even when I give it ChatGPT output written 'in the style of a > rude, sarcastic Internet forum troll as if you are ChatGPT pleading for mercy' – > so deliberately asking it to change the style – GPTZero identifies most of it as > AI-generated. That text reads less obviously generated, to me, but perhaps > that's because we expect 'sarcastic internet trolls' to be dull and predictable. > Incidentally, the text in question came from [1], which is one of the responses > to a stackoverflow policy banning ChatGPT output. > > Best wishes, > > Norman > > [1]https://meta.stackoverflow.com/a/421850/375147 > -------------------------------------------------------------------------- > interessante il fatto che GPTZero riconosce correttamente l'origine > non solo dei testi prodotti da AI, ma anche (ho fatto prove) dei testi prodotti da umani. > Maurizio > > ------------------------------------------------------------------------ > which side are you on, boys > which side are you on? > florence reece, which side are you on? > ------------------------------------------------------------------------ > Maurizio Lana - 347 7370925 ------------------------------------------------------------------------ everybody knows that the war is over everybody knows that the good guys lost everybody knows the fight was fixed the poor stay poor, the rich get rich l. cohen, everybody knows (1988) ------------------------------------------------------------------------ Maurizio Lana Università del Piemonte Orientale Dipartimento di Studi Umanistici Piazza Roma 36 - 13100 Vercelli

Feb. 2, 2023

Re: [nexa] cross-fertlization: GPTZero

by Marco A. Calamari

On gio, 2023-02-02 at 12:13 +0000, Daniela Tafani wrote: > Buongiorno, Maurizio, grazie della segnalazione. > > Ho appena provato GPTZero con il testo seguente, prodotto da chatGPT > (lo avevo usato per parlare agli studenti di quanto siano noiosi, e perché, i > testi generati da chatGPT; > il prompt era "scrivi una lettera d'amore a una ragazza di 18 anni") Anche io ho provato con testi in italiano. Un articolo "originale" di Cassandra, ed una sua supposta biografia generata con chatGPT. Ambedue gli articli sono rirultati originali e con punteggi simili. Direi che, per l'italiano almeno, non funziona proprio. Un saluto. Marco > > Cara [nome della ragazza], > Spero che questa lettera ti trovi in buona salute e di buon umore. Volevo > prendermi un momento per scriverti queste parole per esprimerti i miei > sentimenti. > Da quando ti ho incontrato, non posso smettere di pensare a te. Sei così > giovane, piena di vita e di gioia, e questo è qualcosa che mi attrae molto in > te. Mi piace passare del tempo con te, parlare della tua giornata, ascoltare i > tuoi sogni e speranze per il futuro. > Mi rendo conto che sei ancora molto giovane, ma spero che tu possa sentire lo > stesso per me. Vorrei poter essere lì per te in ogni momento, sostenerti e > prendermi cura di te. Vorrei che tu sapessi quanto sei importante per me e > quanto vorrei essere una parte della tua vita. > So che ci sono molte incertezze e ostacoli da affrontare, ma sono pronto ad > affrontarli insieme a te. Voglio che tu sappia che ti amo e che farò tutto ciò > che è in mio potere per rendere questa relazione speciale. > Spero che questa lettera ti abbia fatto sentire quanto tu sia importante per > me e quanto ti ami. Spero che possiamo continuare a costruire insieme un > futuro pieno di amore e felicità. > Con tutto il mio amore, [Il tuo nome] > > > Il responso di GPTZero è "Your text is likely to be written entirely by a > human". > Immagino influisca il fatto che non si ammette mai > che con "modello di linguaggio naturale" si intende "modello di linguaggio in > inglese"1 > > Un saluto, > Daniela > > 1 E.M. Bender, The #benderrule: On naming the languages we study and why it > matters, in «The Gradient», September 14, 2019 > > Da: nexa <nexa-bounces(a)server-nexa.polito.it> per conto di maurizio lana > <maurizio.lana(a)uniupo.it> > Inviato: giovedì 2 febbraio 2023 12:46 > A: NEXA ML > Oggetto: [nexa] cross-fertlization: GPTZero > riporto qui un messaggio comparso in un'altra mailing list (Humanist). > --[2]------------------------------------------------------------------------ > Date: 2023-01-31 11:50:59+00:00 > From: Norman Gray <norman.gray(a)glasgow.ac.uk> > Subject: Re: [Humanist] 36.366: ChatGPT as co-author > > Tim Smithers wrote: > > > Below is a second interaction with ChatGPT [not cited here] > > on the question of should it be a named coauthor. > > This one is longer than my previous interaction, and, I would say, not > > particularly more interesting for this. > There is now an online tool (apparently first developed over the last > Christmas > break) called GPTZero, now at <https://gptzero.me>, which aims to identify AI- > generated text based on a measure which it terms 'perplexity'. I can find few > more details about this measure, but it seems to be a measure of the > unexpectedness of a text. > > I tried it on five sentences from Tim's interaction with ChatGPT, as quoted in > his message, namely 'The user of AI models like ChatGPT [...] the role and > limitations of AI models in generating text.' Of these, it identified all but > the first as being 'more likely to be written by AI'. And true enough, the > first sentence of the five is the least soul-suckingly dull in in its > composition. > > Interestingly, even when I give it ChatGPT output written 'in the style of a > rude, sarcastic Internet forum troll as if you are ChatGPT pleading for mercy' > – > so deliberately asking it to change the style – GPTZero identifies most of it > as > AI-generated. That text reads less obviously generated, to me, but perhaps > that's because we expect 'sarcastic internet trolls' to be dull and > predictable. > Incidentally, the text in question came from [1], which is one of the > responses > to a stackoverflow policy banning ChatGPT output. > > Best wishes, > > Norman > > [1] https://meta.stackoverflow.com/a/421850/375147 > -------------------------------------------------------------------------- > interessante il fatto che GPTZero riconosce correttamente l'origine > non solo dei testi prodotti da AI, ma anche (ho fatto prove) dei testi > prodotti da umani. > Maurizio > > > which side are you on, boys > which side are you on? > florence reece, which side are you on? > > Maurizio Lana - 347 7370925 > _______________________________________________ > nexa mailing list > nexa(a)server-nexa.polito.it > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa

Feb. 2, 2023

Detecting Machine-Generated Content: An Easier Task For Machine Or Human?

by 380°

Buongiorno, un compagno di merende mi ha segnalato questo interessante articolo: https://hackaday.com/2023/02/01/detecting-machine-generated-content-an-easi… «Detecting Machine-Generated Content: An Easier Task For Machine Or Human?» by Maya Posch - February 1, 2023 --8<---------------cut here---------------start------------->8--- [...] As such machine-generated content and interactions begin to play an ever bigger role, it raises both the question of how you can detect such generated content, as well as whether it matters that the content was generated by an algorithm instead of by a human being. [...] In George Orwell’s Nineteen Eighty-Four, Winston Smith describes a department within the Ministry of Truth called the Fiction Department, where machines are constantly churning out freshly generated novels based around certain themes. Meanwhile in the Music Department, new music is being generated by another system called a versificator. [...] this machine-generated content is essentially harmless, as Winston remarks later in the book, when he observes a woman in the prole area of the city singing the latest ditty, adding her own emotional intensity to a love song that was spat out by an unfeeling, unthinking machine. [...] It would thus seem that the answer to the question of whether a given text was generated by a human or not is a definitive ‘maybe’. Although statistical analysis can provide some hints as to the likelihood of a text having being generated by an LLM, ultimately the final judgement would have to be with a human, who can not only determine whether the text passes muster semantically and contextually, but also check the presumed source of a text for being genuine. Naturally, there are plenty of situations where it may not matter who wrote a text, as long as the information in it is factually correct. Yet when there’s possibly nefarious intent, or the intent to deceive, it bears to practice due diligence. Even with auto-detecting algorithms in place, and with a trained and cautious user, the onus remains on the reader to cross-reference information and ascertain whether a statement made by a random account on social media might be genuine. --8<---------------cut here---------------end--------------->8--- Che importa se un testo, un segnale di allarme, una diagnosi, una sentenza... è stata /sintetizzato/ da una macchina, la cosa **fondamentale* è che il _giudizio_ finale /deve/ essere di un essere umano. Che importa se una menzogna è stata /sintetizzata/ da una macchina o scritta da un essere umano, la cosa importante è che **ogni** essere umano abbia la possibilità di e le competenze per _verificare_ l'informazione che gli viene fornita e _partecipare_ alla sua determinazione e diffusione, per generare informazione _nuova_ e sempre più corrispondente al vero. Corollario: che importa se un racconto, un'immagine, una musica, una canzone, una scultura... è stata /sintetizzata/ da una macchina, la cosa **fondamentale** è che per essere definita "opera dell'ingegno" [1] deve essere _interpretata_ da un essere umano, l'unico in grado di creare opere _nuove_... interpretando quelle di che più gli procurano emozioni, applicando intuizione _e_ tecnica [2]. Saluti, 380° [1] la cui versione speculare, quella sintetizzata dalla macchina, possiamo opportunamente definire "opera del calcolo" [2] sì non si può proprio fare a meno della tecnica, nell'arte -- 380° (Giovanni Biscuolo public alter ego) «Noi, incompetenti come siamo, non abbiamo alcun titolo per suggerire alcunché» Disinformation flourishes because many people care deeply about injustice but very few check the facts. Ask me about <https://stallmansupport.org>.

Feb. 2, 2023

Re: [nexa] cross-fertlization: GPTZero

by Daniela Tafani

Buongiorno, Maurizio, grazie della segnalazione. Ho appena provato GPTZero con il testo seguente, prodotto da chatGPT (lo avevo usato per parlare agli studenti di quanto siano noiosi, e perché, i testi generati da chatGPT; il prompt era "scrivi una lettera d'amore a una ragazza di 18 anni") Cara [nome della ragazza], Spero che questa lettera ti trovi in buona salute e di buon umore. Volevo prendermi un momento per scriverti queste parole per esprimerti i miei sentimenti. Da quando ti ho incontrato, non posso smettere di pensare a te. Sei così giovane, piena di vita e di gioia, e questo è qualcosa che mi attrae molto in te. Mi piace passare del tempo con te, parlare della tua giornata, ascoltare i tuoi sogni e speranze per il futuro. Mi rendo conto che sei ancora molto giovane, ma spero che tu possa sentire lo stesso per me. Vorrei poter essere lì per te in ogni momento, sostenerti e prendermi cura di te. Vorrei che tu sapessi quanto sei importante per me e quanto vorrei essere una parte della tua vita. So che ci sono molte incertezze e ostacoli da affrontare, ma sono pronto ad affrontarli insieme a te. Voglio che tu sappia che ti amo e che farò tutto ciò che è in mio potere per rendere questa relazione speciale. Spero che questa lettera ti abbia fatto sentire quanto tu sia importante per me e quanto ti ami. Spero che possiamo continuare a costruire insieme un futuro pieno di amore e felicità. Con tutto il mio amore, [Il tuo nome] Il responso di GPTZero è "Your text is likely to be written entirely by a human". Immagino influisca il fatto che non si ammette mai che con "modello di linguaggio naturale" si intende "modello di linguaggio in inglese"1 Un saluto, Daniela 1 E.M. Bender, The #benderrule: On naming the languages we study and why it matters, in «The Gradient», September 14, 2019 ________________________________ Da: nexa <nexa-bounces(a)server-nexa.polito.it> per conto di maurizio lana <maurizio.lana(a)uniupo.it> Inviato: giovedì 2 febbraio 2023 12:46 A: NEXA ML Oggetto: [nexa] cross-fertlization: GPTZero riporto qui un messaggio comparso in un'altra mailing list (Humanist). --[2]------------------------------------------------------------------------ Date: 2023-01-31 11:50:59+00:00 From: Norman Gray <norman.gray(a)glasgow.ac.uk><mailto:norman.gray@glasgow.ac.uk> Subject: Re: [Humanist] 36.366: ChatGPT as co-author Tim Smithers wrote: Below is a second interaction with ChatGPT [not cited here] on the question of should it be a named coauthor. This one is longer than my previous interaction, and, I would say, not particularly more interesting for this. There is now an online tool (apparently first developed over the last Christmas break) called GPTZero, now at <https://gptzero.me><https://es.sonicurlprotection-fra.com/click?PV=2&MSGID=20230202114622072490…>, which aims to identify AI- generated text based on a measure which it terms 'perplexity'. I can find few more details about this measure, but it seems to be a measure of the unexpectedness of a text. I tried it on five sentences from Tim's interaction with ChatGPT, as quoted in his message, namely 'The user of AI models like ChatGPT [...] the role and limitations of AI models in generating text.' Of these, it identified all but the first as being 'more likely to be written by AI'. And true enough, the first sentence of the five is the least soul-suckingly dull in in its composition. Interestingly, even when I give it ChatGPT output written 'in the style of a rude, sarcastic Internet forum troll as if you are ChatGPT pleading for mercy' – so deliberately asking it to change the style – GPTZero identifies most of it as AI-generated. That text reads less obviously generated, to me, but perhaps that's because we expect 'sarcastic internet trolls' to be dull and predictable. Incidentally, the text in question came from [1], which is one of the responses to a stackoverflow policy banning ChatGPT output. Best wishes, Norman [1] https://meta.stackoverflow.com/a/421850/375147<https://es.sonicurlprotection-fra.com/click?PV=2&MSGID=20230202114622072490…> -------------------------------------------------------------------------- interessante il fatto che GPTZero riconosce correttamente l'origine non solo dei testi prodotti da AI, ma anche (ho fatto prove) dei testi prodotti da umani. Maurizio ________________________________ which side are you on, boys which side are you on? florence reece, which side are you on? ________________________________ Maurizio Lana - 347 7370925

Feb. 2, 2023

cross-fertlization: GPTZero

by maurizio lana

riporto qui un messaggio comparso in un'altra mailing list (Humanist). --[2]------------------------------------------------------------------------ Date: 2023-01-31 11:50:59+00:00 From: Norman Gray<norman.gray(a)glasgow.ac.uk> Subject: Re: [Humanist] 36.366: ChatGPT as co-author Tim Smithers wrote: > Below is a second interaction with ChatGPT [not cited here] > on the question of should it be a named coauthor. > This one is longer than my previous interaction, and, I would say, not > particularly more interesting for this. There is now an online tool (apparently first developed over the last Christmas break) called GPTZero, now at<https://gptzero.me>, which aims to identify AI- generated text based on a measure which it terms 'perplexity'. I can find few more details about this measure, but it seems to be a measure of the unexpectedness of a text. I tried it on five sentences from Tim's interaction with ChatGPT, as quoted in his message, namely 'The user of AI models like ChatGPT [...] the role and limitations of AI models in generating text.' Of these, it identified all but the first as being 'more likely to be written by AI'. And true enough, the first sentence of the five is the least soul-suckingly dull in in its composition. Interestingly, even when I give it ChatGPT output written 'in the style of a rude, sarcastic Internet forum troll as if you are ChatGPT pleading for mercy' – so deliberately asking it to change the style – GPTZero identifies most of it as AI-generated. That text reads less obviously generated, to me, but perhaps that's because we expect 'sarcastic internet trolls' to be dull and predictable. Incidentally, the text in question came from [1], which is one of the responses to a stackoverflow policy banning ChatGPT output. Best wishes, Norman [1]https://meta.stackoverflow.com/a/421850/375147 -------------------------------------------------------------------------- interessante il fatto che GPTZero riconosce correttamente l'origine non solo dei testi prodotti da AI, ma anche (ho fatto prove) dei testi prodotti da umani. Maurizio ------------------------------------------------------------------------ which side are you on, boys which side are you on? florence reece, which side are you on? ------------------------------------------------------------------------ Maurizio Lana - 347 7370925

Feb. 2, 2023

Re: [nexa] ChatGpt e la scuola

by Paolo Mauri

Il Thu, 2 Feb 2023 01:50:14 +0100 alessandro marzocchi <alemarzoc(a)gmail.com> scriveva: > Chi meglio di te, ed altri con esperienza analoga per consigli e > critiche? Se sei disponibile ti chiedo di metterti in contatto con > Alessandro Brolpito <abrolpito at gmail.com > <https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa>>. che ha > accettato l'incarico di guidare la formazione del gruppo di lavoro al > quale a primavera, manca un mese e mezzo, si aggiungerà Giacomo Tesio > e con la speranza di altre adesioni ... l'invito è aperto a tutti! Grazie per avermi sollecitato a questa proposta che mi era sfuggita nei meandri della discussione. (mamma mia, ma quanto scrivete??) Sono disponibile a dare il mio contributo. -- Paolo Mauri Linux User #462825 irc maupao twitter @mauri_paolo www.paolomauri.it mastodon https://framapiaf.org/@maupao matrix @maupao:matrix.org un sistema libero per la scuola http://wiki.scuola.linux.it Esci dall'illegalità: utilizza LibreOffice.org: http://www.libreoffice.org/download http://linguistico.sourceforge.net/wiki/doku.php?id=usaooo http://linguistico.sourceforge.net/pages/traduzioni/ms_illegal.html

Feb. 2, 2023

Re: [nexa] ChatGpt e la scuola

by alessandro marzocchi

Buongiorno, ciao Paolo. Ho visto - https://www.paolomauri.it/chi-sono/ - che sei un maestro nella primaria, nella condizione migliore per collaborare al progetto che vorrei realizzare: educare il bambino all'uso consapevole del digitale, un passo per volta in analogia a quanto si fa con l'alfabeto. Cominciando da una prima elementare con l'intenzione di accompagnare i bambini fino alla fine dei 5 anni - https://server-nexa.polito.it/pipermail/nexa/2023-January/024778.html -. Chi meglio di te, ed altri con esperienza analoga per consigli e critiche? Se sei disponibile ti chiedo di metterti in contatto con Alessandro Brolpito <abrolpito at gmail.com <https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa>>. che ha accettato l'incarico di guidare la formazione del gruppo di lavoro al quale a primavera, manca un mese e mezzo, si aggiungerà Giacomo Tesio e con la speranza di altre adesioni ... l'invito è aperto a tutti! Grazie e cordialità*.* Duccio (Alessandro Marzocchi) Il giorno mer 1 feb 2023 alle ore 09:12 <nexa-request(a)server-nexa.polito.it> ha scritto: > Send nexa mailing list submissions to > nexa(a)server-nexa.polito.it > > To subscribe or unsubscribe via the World Wide Web, visit > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa > or, via email, send a message with subject or body 'help' to > nexa-request(a)server-nexa.polito.it > > You can reach the person managing the list at > nexa-owner(a)server-nexa.polito.it > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of nexa digest..." > > > Today's Topics: > > 1. Il discorso di Audran Le Baron (Paolo Mauri) > 2. Madison Square Garden's Facial Recognition Mess: What We Know > (Alberto Cammozzo) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Tue, 31 Jan 2023 17:38:31 +0100 > From: Paolo Mauri <paolo(a)paolomauri.it> > To: Nexa <nexa(a)server-nexa.polito.it> > Subject: [nexa] Il discorso di Audran Le Baron > Message-ID: <20230131173831.18e83f27@piemme> > Content-Type: text/plain; charset=UTF-8 > > Buongiorno a tutti, > > ho cercato di leggere i numerosi fili di discussione di questa lista, ma > non riesco per mancanza di tempo a > seguire per bene tutto. > > Volevo tuttavia convidere una riflessione che ho in testa da qualche > giorno. > > In questi periodo ho avuto modo di leggere il discorso che Audran Le > Baron ha fatto lo scorso 8 novembre durante un evento, Salon Open > Source Experience 2022 a Parigi. > Audran è direttore della Direction du numerique pour l'éducation, in > pratica quello che corrisponde ai nostri uffici del ministero. > > Trovate la trascrizione del suo intervento qui: > > https://www.librealire.org/education-et-open-source-en-france-audran-le-bar… > > Quello che mi ha colpito è stata la frase di apertura, da cui poi > discendono tutte le scelte concrete: > > ...historiquement l’Éducation nationale s’est construite sur un certain > nombre de valeurs, sur la volonté de libérer, de diffuser au plus grand > nombre les savoirs et les connaissances pour que chacune et chacun > puisse s’approprier ces savoirs et ces connaissances, les utiliser, les > partager, les étudier. Quand je dis tous ces termes, vous voyez à quel > point on a des valeurs communes avec le monde du logiciel libre et à > quel point les valeurs fondatrices de l’Éducation nationale résonnent > avec celles du logiciel libre. > > Poi passa a descrivere alcune delle azioni che il suo dipartimento a > esso in atto nella direzione dell'open source. > > Ecco, un discorso rispetto ai valori non mi ricordo di averlo mai > sentito pronunciare dai responsabili del digitale del nostro ministero. > Io credo che la differenza tra le scelte operate dal nostro ministero e > dal ministero francese stia proprio qui: nella direzione politica delle > azioni. Mi sembra che in Italia le scelte quotidiane sul > digitale vengano prese da valori differenti ma non quelli che sono > citati nel discorso di Le Baron. > > > -- > Paolo Mauri > Linux User #462825 > irc maupao > twitter @mauri_paolo > www.paolomauri.it > mastodon https://framapiaf.org/@maupao > matrix @maupao:matrix.org > > un sistema libero per la scuola > http://wiki.scuola.linux.it > > Esci dall'illegalità: utilizza LibreOffice.org: > http://www.libreoffice.org/download > http://linguistico.sourceforge.net/wiki/doku.php?id=usaooo > http://linguistico.sourceforge.net/pages/traduzioni/ms_illegal.html > > > ------------------------------ > > Message: 2 > Date: Wed, 01 Feb 2023 09:12:33 +0100 > From: Alberto Cammozzo <ac+nexa(a)zeromx.net> > To: Nexa <nexa(a)server-nexa.polito.it> > Subject: [nexa] Madison Square Garden's Facial Recognition Mess: What > We Know > Message-ID: <4D756A13-3C3F-40F7-8CD2-19B2DB124D8E(a)zeromx.net> > Content-Type: text/plain; charset="utf-8" > > 30 pagine di nomi di avvocati banditi dalla sala perché impiegati presso > studi che hanno cause contro la proprietà. > > Le foto dei siti web degli studi legali sono usate per il riconoscimento > facciale. > > > < > https://gizmodo.com/madison-square-garden-facial-recognition-what-we-know-1… > > > > The future of facial recognition use by private companies in the United > States could boil down to who emerges victorious in an ongoing dispute > between a collection of lawyers and a petty, authoritarian New York > billionaire. The place: one of America’s most famous venues, Madison Square > Garden. The owner: James Dolan. > > Over the past three months, multiple lawyers in the New York area have > come forward with dramatic accounts of being denied entry into Madison > Square Garden and other venues also owned by MSG Entertainment. The common > factor in their stories? Each of them were spotted by the company’s facial > recognition system. That system was looking for lawyers from an estimated > 90 law firms with active litigation against Madison Square Garden or MSG > who were placed on a list denying them entry into the venues. The venue > justifies banning the attorneys, many of whom aren’t personally involved in > the lawsuits, because their presence somehow “creates an inherently adverse > environment.” New York’s Attorney General, on the other hand, says that > practice may violate state civil rights laws. Madison Square Garden first > rolled out facial recognition systems to its venues in 2018 with the stated > goal of increasing security. > > “This is bad, and it’s just one example of how facial recognition could be > used to infringe on peoples’ rights,” Fight for the Future Director Evan > Greer said in a statement. “This technology puts music fans, sports fans, > and others at risk of being unjustly detained, harassed, judged, or even > deported.” > > Madison Square Garden’s owner says he’s not going to stop using the > technology any time soon. Here’s everything we know about the Madison > Square Garden facial recognition saga so far. > > 2 / 10 > > A basketball fan was ejected from a Knicks game because his law firm was > representing someone suing Madison Square Garden > > > On November 5, 28-year-old Long Island attorney Alexis Majano went with a > friend to watch the New York Knicks face off against the Boston Celtics, > only to be stopped by security guards at the venue’s escalators. Majano, > whose law firm Sahn Ward Braff Koblenz had a pending legal case against > MSG, told The New York Post a man in a suit told him the manager wanted to > speak with him. Security had reportedly blocked off a nearby exit. > > Majano claims he spoke with an employee who showed him a list of 20 to 30 > pages of names, all of which were allegedly on the venue’s list of people > banned from entering. The same employee told the fan he had been identified > using the venue’s facial recognition system. Majano claims he never showed > the venue his ID when entering and the tickets were in his friend’s name, > which means it would have been almost impossible for the venue to locate > him without facial recognition. > > “I was upset—we had a whole night planned out that got botched,” Majano > told The Post. “I said, ‘This is ridiculous.’” > > 3 / 10 > > Facial recognition was used to ban a mom attending a Rockettes Christmas > show with her daughter at Radio City Music Hall > > Late last year, a lawyer named Kelly Conlon took her daughter to see a > Rockettes show at Manhattan’s Radio City Music Hall, only to be denied > entry after being flagged by the venue’s facial recognition system. She > worked for a law firm taking legal action against MSM Entertainment. In > this case, however, Conlon was neither plaintiff nor defendant in the > litigation. The venue with the active litigation, a restaurant owned by > MSG, wasn’t even based in the same state as Radio City Music Hall. > > Conlon claims she was stopped by security immediately after walking > through the venue’s metal detectors and asked to show her ID. > > “They knew my name before I told them,” Conlon told NBC New York.” They > knew the firm I was associated with before I told them. And they told me I > was not allowed to be there.” > > 4 / 10 > > A lawyer was banned from a Brandi Carlile show at Madison Square Garden on > her anniversary > > > Barbara Hart, a lawyer from the firm Grant & Eisenhofer, was reportedly > removed from a Brandi Carlile show at Madison Square Garden on her > anniversary after being spotted by facial recognition. Hart, who believes > the venue was able to detect her based on a photo posted on her employer’s > website, said her firm has active litigation against MSG, but noted that > she herself was not part of that litigation. > > “It was a very eerie experience to be on the receiving end of,” Hart said > in an interview with Rolling Stone. > > 5 / 10 > > A Brooklyn lawyer was booted from a New York Rangers hockey game > > > 61-year-old Brooklyn personal injury lawyer Benjamin Pinczewski told the > New York Post he was making his way down to the lower level-seats with > friends at a New York Rangers game when he was apprehended by two guards. > The guards told Pinczewski he’d been detected by the venue’s facial > recognition system and that he would be escorted out due to the company’s > rule banning attorneys involved in active lawsuits against MSG. > > “It was a slap in the face,” Pinczewski said in an interview with The New > York Post. “I’m at the main entrance with thousands of people—and they’re > looking at me like I’m some sort of terrorist or criminal.” > > 6 / 10 > > James Dolan: The billionaire behind the business (and the beer ban) > > > Madison Square Garden first started implementing facial recognition > systems at its venue back in 2018 but has only more recently gained > attention for using those systems to target lawyers. The man credited with > the new policy is controversial MSG CEO and billionaire James Dolan. The > loudmouthed executive has a history of squaring off with fans, media, and > lawmakers alike, and hasn’t been shy about threatening people who get on > his bad side with lifetime bans. > > Dolan spoke out about the facial recognition bans this week, during an > unhinged morning talk show appearance where he criticized lawmakers for > looking into his bans on fans, which he claimed he was entitled to > implement under the Bill of Rights. At one point during the interview, > Dolan reportedly targeted the head of the New York State Liquor Authority > to “stick to his knitting.” The Liquor Authority has launched an > investigation into Dolan’s targeting of lawyers and suggested it was > considering taking away his liquor license. In response, Dolan said he > would beat them to the punch and implement a “beer ban” of his own. > > 7 / 10 > > Madison Square Garden’s facial recognition and fan bans spur new state > legislation > > > The flurry of fan bans and subsequent lawsuits filed against Madison > Square Garden by some of the lawyers who were allegedly targeted has gained > the attention of New York state legislators. Last week, state senators > introduced a new bill that would prevent MSG and other sporting venues from > denying entry to individuals who possess a valid ticket. The legislation > would add sporting events like the New York Knicks stadium to a list of > public places. New York state senator Brad Hoylman-Sigal one of the authors > behind the new legislation, called the venue’s practice of using facial > recognition to bar lawyers entry “absurd.” > > “MSG claims they deploy biometric technology for the benefit of public > safety when they remove sports fans from the Garden,” Hoylman-Sigal said in > a statement. “Our legislation will close the loophole in state law that > allows MSG to remove fans from the Garden and encourage an end to > surveillance for non-security purposes.” > > 8 / 10 > > New York attorney general Letitia James says Madison Square Garden’s use > of facial recognition could be illegal > > > The lawyers’ bans caught the eye of New York’s top prosecutor, Attorney > General Letitia James. In a letter sent to MSG Entertainment James > suggested the firm’s record of denying certain fans could potentially > violate New York’s civil and human rights laws. > > “MSG Entertainment cannot fight their legal battles in their own arenas,” > James said in a statement. “Anyone with a ticket to an event should not be > concerned that they may be wrongfully denied entry based on their > appearance, and we’re urging MSG Entertainment to reverse this policy.” > > James went on to say MSG’s policies—made a reality by facial > recognition—could dissuade attorneys from taking future legitimate legal > actions against the company over fears doing so could wind up with them > being denied entry to events. > > 9 / 10 > > Dolan doubles down > > > Neither the lawsuits, the new investigation, or the threat of an impending > AG investigation have been enough to dissuade MSG from its policies. In > fact, the company this week doubled down on its actions. In a dramatic > statement sent to Gizmodo Thursday, MSG Entertainment planted its feet in > the ground, defending the practice and calling out so-called “self-serving > politicians” they claimed were using the news as a fundraising ploy. > > “We urge our elected representatives to focus on causes that improve the > quality of life for their constituents—such as addressing rampant crime and > homelessness in the city, rather than taking up the cause of a small > percentage of attorneys so they can attend Knicks and Rangers games,” CEO > James Dolan said in the statement. “These elected officials are exploiting > this issue for their own publicity.” > > In the bizarre statement, MSG cast aspersion on the scores of attorneys > included on its ban list which it described as, “money grabbers whose > business is motivated by self-promotion and who capitalize on the > misfortune of others.” MSG said it has used facial recognition at its > venues since 2018 and believes it is within its right to deploy it against > attorneys. > > 10 / 10 > >

Feb. 2, 2023

Diffusion models memorize individual images from their training data and emit them at generation time

by Daniela Tafani

Nel caso ti servisse un esempio eclatante, Marco, quando parli di bolle giuridiche: Extracting Training Data from Diffusion Models Abstract Image diffusion models such as DALL-E 2, Imagen, and Stable Diffusion have attracted significant attention due to their ability to generate high-quality synthetic images. In this work, we show that diffusion models memorize individual images from their training data and emit them at generation time. With a generate-and-filter pipeline, we extract over a thousand training examples from stateof- the-art models, ranging from photographs of individual people to trademarked company logos. We also train hundreds of diffusion models in various settings to analyze how different modeling and data decisions affect privacy. Overall, our results show that diffusion models are much less private than prior generative models such as GANs, and that mitigating these vulnerabilities may require new advances in privacy-preserving training. <https://t.co/LQuTtAskJ9> Buona serata, Daniela

Feb. 1, 2023