nexa
By thread
nexa@server-nexa.polito.it
By month
Messages by month
- ----- 2026 -----
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2025 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2024 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2023 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2022 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2021 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2020 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2019 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2018 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2017 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2016 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2015 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2014 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2013 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2012 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2011 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2010 -----
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- ----- 2009 -----
- December
- November
- October
- September
- August
- July
- June
- May
- 40 participants
- 30621 messages
Articolo su algoretica
by Guido Boella
Carissime e carissimi
segnalo un articolo sul tema algoretica scritto da SIpEIA su Magazine
Intelligenza Artificiale - MagIA
https://magia.news/che-dire-della-algor-etica-spunti-per-una-critica-della-…
colgo l'occasione per ricordare che MagIA vuole essere un luogo aperto a
tutte e tutti per parlare di impatto dell'IA e quindi rinnovo la call to
action a tutte e tutti gli interessati a contribuire con editoriali se
interessate/i
saluti
guido
--
Dipartimento di Informatica
Università di Torino
Cso Svizzera 185
10149 Torino
Vice-Rettore Vicario dell'Università di Torino per la promozione dei
rapporti con le imprese e le associazioni di categoria delle imprese e per
il coordinamento con le iniziative di innovazione industriale sul territorio
Vicepresidente del Competence Center CIM4.0 www.cim40.com
Visitate i siti delle nostre iniziative
www.magia.news
www.firstlife.org
www.commonshood.eu
www.projectco3.eu
www.sipeia.it
www.chedih.eu
www.ai-aware.eu
nlab4cit.eu
June 21, 2024
Re: [nexa] EU cancels vote on child sexual abuse law amid encryption concerns
by Andrea Barontini
Una temporanea buona notizia, ci voleva
Grazie della segnalazione
Il 20/06/24 13:43, Daniela Tafani ha scritto:
> EU cancels vote on child sexual abuse law amid encryption concerns
>
> Countries clash over how to safeguard privacy when rooting out illegal pictures and grooming on private chat apps.
>
> une 20, 2024 12:29 pm CET
> By Clothilde Goujard
>
> A vote scheduled today to amend a draft law that may require WhatsApp and Signal to scan people’s pictures and links for potential child sexual abuse material was removed from European Union countries' agenda, according to three EU diplomats.
>
> Ambassadors in the EU Council were scheduled to decide whether to back a joint position on an EU regulation to fight child sexual abuse material (CSAM). But many EU countries including Germany, Austria, Poland, the Netherlands and the Czech Republic were expected to abstain or oppose the law over cybersecurity and privacy concerns.
>
> "In the last hours, it appeared that the required qualified majority would just not be met," said an EU diplomat from the Belgian presidency, which is spearheading negotiations until end June as chair of the EU Council.
>
> The draft law, proposed in 2022, has drawn controversy for potentially forcing messaging apps to scan all images and links to find and report child abuse material and conversations between potential offenders and minors, known as grooming. Privacy groups have cried foul over the law, saying it effectively breaks end-to-end encrypted messaging.
>
> European Commission Vice President Věra Jourová said Thursday the Commission's original proposal meant “that even encrypted messaging can be broken for the better protection of children.”
>
> The Belgian Council presidency has been trying for the last six months to solve a deadlock among EU countries to move negotiations forward to finalize the law.
>
> Some EU heavyweights like Germany and Poland have backed privacy experts' warnings that it threatens privacy. Others like Ireland and Spain have insisted on the need for a strong law to monitor online content amid a spike in child sexual abuse material since the pandemic.
>
> Under the Belgians' plan, obtained by POLITICO earlier, messaging apps would scan pictures and links when users upload them via their services, and users would be informed of this under the terms and conditions. Users who refused the regime would be blocked from sending pictures and links.
>
> Highly secure apps using end-to-end encryption like WhatsApp, Signal and Messenger would also have to respect such measures. The draft proposal however exempted “accounts used by the State for national security purposes."
>
> Once EU countries agree on a joint position, they will still have to negotiate the final version of the law with the European Parliament and European Commission. Parliament has taken a more privacy-friendly stance in its own version of the law adopted in November 2023.
>
>
> https://www.politico.eu/article/eu-council-cancels-vote-on-encryption-break…
June 20, 2024
EU cancels vote on child sexual abuse law amid encryption concerns
by Daniela Tafani
EU cancels vote on child sexual abuse law amid encryption concerns
Countries clash over how to safeguard privacy when rooting out illegal pictures and grooming on private chat apps.
une 20, 2024 12:29 pm CET
By Clothilde Goujard
A vote scheduled today to amend a draft law that may require WhatsApp and Signal to scan people’s pictures and links for potential child sexual abuse material was removed from European Union countries' agenda, according to three EU diplomats.
Ambassadors in the EU Council were scheduled to decide whether to back a joint position on an EU regulation to fight child sexual abuse material (CSAM). But many EU countries including Germany, Austria, Poland, the Netherlands and the Czech Republic were expected to abstain or oppose the law over cybersecurity and privacy concerns.
"In the last hours, it appeared that the required qualified majority would just not be met," said an EU diplomat from the Belgian presidency, which is spearheading negotiations until end June as chair of the EU Council.
The draft law, proposed in 2022, has drawn controversy for potentially forcing messaging apps to scan all images and links to find and report child abuse material and conversations between potential offenders and minors, known as grooming. Privacy groups have cried foul over the law, saying it effectively breaks end-to-end encrypted messaging.
European Commission Vice President Věra Jourová said Thursday the Commission's original proposal meant “that even encrypted messaging can be broken for the better protection of children.”
The Belgian Council presidency has been trying for the last six months to solve a deadlock among EU countries to move negotiations forward to finalize the law.
Some EU heavyweights like Germany and Poland have backed privacy experts' warnings that it threatens privacy. Others like Ireland and Spain have insisted on the need for a strong law to monitor online content amid a spike in child sexual abuse material since the pandemic.
Under the Belgians' plan, obtained by POLITICO earlier, messaging apps would scan pictures and links when users upload them via their services, and users would be informed of this under the terms and conditions. Users who refused the regime would be blocked from sending pictures and links.
Highly secure apps using end-to-end encryption like WhatsApp, Signal and Messenger would also have to respect such measures. The draft proposal however exempted “accounts used by the State for national security purposes."
Once EU countries agree on a joint position, they will still have to negotiate the final version of the law with the European Parliament and European Commission. Parliament has taken a more privacy-friendly stance in its own version of the law adopted in November 2023.
https://www.politico.eu/article/eu-council-cancels-vote-on-encryption-break…
June 20, 2024
Re: [nexa] "ChatGPT is bullshit"
by benjuade
E collegato a ciò:
Https://mastodon.social/@smaurizi/112648350606223172
nel silenzio e nella totale assenza di dibattito sulle implicazioni, il governo ci ha messo sotto il giogo dalla #NSA, con la vendita della rete #TIM al fondo #KKR, il cui partner è l'ex capo della #CIA, Petraeus
On 20 June 2024 00:48:09 CEST, Maria Chiara Pievatolo <mariachiara.pievatolo(a)unipi.it> wrote:
>On 19/06/24 18:38, Enrico Nardelli wrote:
>> Grazie Daniela per questo riferimento.
>> La citazione ricordata da Ezra Klein nell'intervista a Gary Marcus, sul fatto che una cosa può essere falsa ma, a parte il resto, non inferiore alla cosa reale, mi ha fatto venire in mente l'articolo di Daniel Dennet sul fatto che generare finzione attraverso l'IA è un atto immorale come produrre denaro falso.
>
>E che però rimarrà assai utile, anche quando la bolla scoppierà, e non solo per giocare a monopoli(i), come ha fatto notare Smowden a proposito del recente ingresso nel consiglio di amministrazione di OpenAI del generale in pensione Paul Nakasone "former National Security Administration (NSA) Director, who also helmed the military's cybersecurity-focused Cyber Command unit".
>
>"They've gone full mask off: do not ever trust OpenAI or its products," Snowden — emphasis his — wrote in a Friday post to X-formerly-Twitter, adding that "there's only one reason for appointing" an NSA director "to your board."
>
>"This is a willful, calculated betrayal of the rights of every person on earth," he continued. "You've been warned."
>
>https://futurism.com/the-byte/snowden-openai-calculated-betrayal
>
>E buonanotte,
>MCP
June 20, 2024
Re: [nexa] "ChatGPT is bullshit"
by 380°
Maria Chiara Pievatolo <mariachiara.pievatolo(a)unipi.it> writes:
[...]
> come ha fatto notare Smowden a proposito del recente ingresso nel
> consiglio di amministrazione di OpenAI del generale in pensione Paul
> Nakasone "former National Security Administration (NSA) Director, who
> also helmed the military's cybersecurity-focused Cyber Command unit".
[...]
> https://futurism.com/the-byte/snowden-openai-calculated-betrayal
con una, spero inutile, precisazione: che questo tipo di /commistione/
non è affatto nuova, anzi!
si tratta "semplicemente" di riconoscere che viviamo nel
"military–industrial complex" del quale ci avvisò per la prima (?) volta
Dwight D. Eisenhower nel 1961 nel suo Farewell Address to the Nation
[1]
https://en.wikipedia.org/wiki/Military%E2%80%93industrial_complex#Etymology
...mica un _complottista_ qualsiasi e _mica_ un antimilitarista
qualsiasi e... diciamo una persona _molto_ ben informata sui fatti.
«Yet we must not fail to comprehend its grave implications.»
alcune /congetture/ attorno a quel discorso sostengono che il termine
preciso nelle bozze fosse:
1. war-based industrial complex
2. military–industrial–congressional complex
3. military–industrial–scientific complex
4. military–industrial–academic complex
come potete facilmente intuire è difficile che dal "complex" sia escluso
il controllo della ricerca scientifica e delle applicazioni
tecnologiche, Internet e AI inclusa ovviamente
Attorno al concetto molti intellettuali si sono cimentati in studi,
ricerche e interpretazioni, tra cui trovo interessante:
--8<---------------cut here---------------start------------->8---
Norman Cousins's 1987 book The Pathology of Power, "Were the Soviet
Union to sink tomorrow under the waters of the ocean, the American
military–industrial complex would have to remain, substantially
unchanged, until some other adversary could be invented. Anything else
would be an unacceptable shock to the American economy."
--8<---------------cut here---------------end--------------->8---
fermiamoci qui :-)
saluti, 380°
[1] video integrale: https://www.c-span.org/video/?15026-1/president-dwight-eisenhower-farewell-…
--
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>.
June 20, 2024
Re: [nexa] "ChatGPT is bullshit"
by 380°
Buongiorno,
OK, entriamo nella stalla del toro :-)
Daniela Tafani <daniela.tafani(a)unipi.it> writes:
> Il 19/06/2024 18:38, Enrico Nardelli ha scritto:
[....]
>> Daniel Dennet sul fatto che generare finzione attraverso l'IA
per la precisione: generare identità contraffatte (si veda sotto), non
"finzione" in generale
>> è un atto immorale come produrre denaro falso.
innanzitutto, sapendo come e perché viene prodotto (e usato) il denaro
/vero/ - mi sto riferendo /esattamente/ al combinato disposto del
"military–coinage–slave complex" postulato da (l'antropologo) David
Graeber [1] e la quantità impressionante di debito _creato_ dalla FED
negli ultimi anni - devo confessare che ho sviluppato un discreto
rispetto per i falsari di denaro, rispetto che cresce in modo
inversamente proporzionale a quello che ho per i falsari della verità
poi: è _immorale_ solo "generare finzione" (ovviamente senza
dichiararlo) /attraverso l'AI/?
quante "cose" che ora crediamo _vere_ (o _false_ [2]) sono finzioni ben
congeniate da menti umane che un grande uomo definì "raffinatissime"?
l'ho già detto: io sono _felice_ che l'uso dell'IA stia /facilitando/ la
produzione di "denaro falso", così tocca imparare, faticosamente, a
riconoscerlo :-D
tipo riconoscere le debolezze del Turing Test quando si tratta di
_discriminare_ se dall'altra parte c'è una persona o un _software_:
https://en.wikipedia.org/wiki/Turing_test#Weaknesses
--8<---------------cut here---------------start------------->8---
Numerous experts in the field, including cognitive scientist Gary
Marcus, insist that the Turing test only shows how easy it is to fool
humans and is not an indication of machine intelligence.
[...] To successfully appear human, there is no need for the machine to
have any intelligence whatsoever and only a superficial resemblance to
human behaviour is required.
[...] Since 2004, the Loebner Prize organisers have deployed
philosophers, computer scientists, and journalists among the
interrogators. Nonetheless, some of these experts have been deceived by
the machines.
[...] Since human behaviour and intelligent behaviour are not exactly
the same thing, the test can fail
[...] The Turing test is concerned strictly with how the subject acts –
the external behaviour of the machine. In this regard, it takes a
behaviourist or functionalist approach to the study of the mind.
--8<---------------cut here---------------end--------------->8---
e mi fermo qui: "behaviourist or functionalist approach" bastano per
evidenziare la puzza.
eh sì, _queste_ cose dovrebbero insegnarle a partire dalle scuole
elementari, altro che "nativi digitali"
...eh certo che però questo implicherebbe per forza di cose il
miglioramento collettivo della capacità di distinguere le stronzate dai
semplici errori, e questo per chi _amministra_ è un serio problema.
>> https://www.theatlantic.com/technology/archive/2023/05/problem-counterfeit-…
ecco cosa dice Dennet nel suo articolo:
--8<---------------cut here---------------start------------->8---
The Problem With Counterfeit People
Companies using AI to generate fake people are committing an immoral act
of vandalism, and should be held liable.
[...] Today, for the first time in history, thanks to artificial
intelligence, it is possible for anybody to make counterfeit people who
can pass for real in many of the new digital environments we have
created. These counterfeit people are the most dangerous artifacts in
human history, capable of destroying not just economies but human
freedom itself. Before it’s too late (it may well be too late already)
we must outlaw both the creation of counterfeit people and the “passing
along” of counterfeit people. The penalties for either offense should be
extremely severe, given that civilization itself is at risk.
[...] The philosopher and historian Yuval Noah Harari, writing in The
Economist in April, ended his timely warning about AI’s imminent threat
to human civilization with these words:
“This text has been generated by a human. Or has it?”
[...] The key design innovation in the technology that makes losing
control of these systems a real possibility is that, unlike nuclear
bombs, these weapons can reproduce. Evolution is not restricted to
living organisms, as Richard Dawkins demonstrated in 1976 in The Selfish
Gene.
[...] They [AI generated Counterfei people] will learn from one another,
and those that are the smartest, the fittest, will not just survive;
they will multiply.
--8<---------------cut here---------------end--------------->8---
l'articolo quindi, da quello che capisco, si riferisce alla
falsificazione di identità, non alle identità fittizie (pseudonime),
corretto? Nemmeno al "generare finzione" in generale, giusto?
cioè: io sono "tua madre", "il tuo capo ufficio", ecc....
da quello che so la falsificazione dell'identità è già reato in diversi
ordinamenti: è il caso di farne una apposta se la falsificazione è
ottenuta per mezzo di "IA"?
e ancora: ogni volta che qualcuno cita Yuval Noah Harari ad-minchiam per
avvalorare una tesi, la mia intelligenza /emotiva/ mi spinge a cestinare
l'intero ragionamento
e ancora: parlare di "AI" (quindi di software) in termini di "weapons" è
già sufficientemente fuorviante da sé, ma se in più si /suggerisce/ la
_congettura_ che l'evoluzione _darwiniana_ appartenga anche a "non
living organisms" e che l'abbia _dimostrato_ (eh?!?!?) Richard Dawkins
[3] e _quindi_ sistemi AI creeranno persone false che si /riprodurranno/
e diventeranno sempre più "smart and fit": ma che davero!??!?
è davvero il caso di ingaggiare una simile dialettica per contrastare
gli effetti negativi delle stronzate generate dall'IA?
faccio mie le parole dell'ultimo paragrafo dell'articolo di Dennet:
--8<---------------cut here---------------start------------->8---
[...] Another pandemic is coming, this time attacking the fragile
control systems in our brains—namely, our capacity to reason with one
another—that we have used so effectively to keep ourselves relatively
safe in recent centuries.
--8<---------------cut here---------------end--------------->8---
con una precisazione: _questo_ tipo di "pandemia", quella che inficia la
nostra capacità di ragionare assieme, è _endemica_ da secoli, altro che
visione nostalgico-romantica dei "bei tempi che furono"! (non c'è
bisogno che sia io a ricordarvi l'obnubilazione della ragione a cui
abbiamo /storicamente/ "assistito" in "recent centuries")
[...]
saluti, 380°
[1] «Debt: The First 5000 Years»
https://en.wikipedia.org/wiki/Debt:_The_First_5000_Years (n.d.r.:
sull'abolizione della schiavitu ci sarebbe /qualcosa/ da dire, ma...)
[2] tipo il terrapiattismo, che è talmente ovvio che è falso che puzza
lontano anni luce di "false flag"
[3] https://en.wikipedia.org/wiki/The_Selfish_Gene
P.S.: poi, dopo il reato di falsificazione di identità umana _da parte_
di "una AI (di decima generazione)" introdurremo anche il reato di
falsificazione di identità AI _da parte_ di una persona:
https://en.wikipedia.org/wiki/Confederate_effect? :-D
--
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>.
June 20, 2024
Re: [nexa] "ChatGPT is bullshit"
by Giacomo Tesio
Salve Daniela,
Il 19 Giugno 2024 11:15:55 UTC, Daniela Tafani <daniela.tafani(a)unipi.it> ha scritto:
>
> GARY MARCUS: It’s exactly right. These systems have no conception of truth. Sometimes
> they land on it and sometimes they don’t, but they’re all fundamentally bullshitting in the sense that
> they’re just saying stuff that other people have said and trying to maximize the probability of that.
> It’s just auto complete, and auto complete just gives you bullshit.
Sono noioso, ma non posso fare a meno di sottolineare come, anche in critiche come
questa, il soggetto delle frasi sposta l'attenzione sullo strumento, sollevando chi lo
amministra da qualsiasi responsabilità.
È ovvio [1] che l'output di questi software programmati statisticamente non
può avere alcuna relazione con la verità.
È ovvio che nessun software ha conoscenza, né del bene o del male, né del vero o del falso.
È ovvio che nessun software può avere intenzionalità propria per cui non può
scegliere di mentire o dire a verità.
Ma non è ChatGPT a distribuire sul web i propri output!
L'intenzionalità c'è ed è di OpenAI, Microsoft etc...
Distribuendo online l'output del software
che hanno scritto ed amministrano, queste società scelgono di diffondere testi che non
hanno corrispondenza con la realtà.
Dunque, quandanche la tipologia di software di cui parliamo fosse qualificabile come
bullshit generator, gli output diffusi da OpenAI, Microsoft, Google etc.. sono semplicemente
disinformazione.
Il fatto che ogni tanto tali output possano essere interpretati come informazioni corispondenti
alla realtà, serve esattamente allo scopo di diffondere disinformazione: come
sappiamo tutti, per rendere una menzogna più credibile e perniciosa, è utile diffomderla
insieme a tante verità note o facilmente verificabili (magari con una ricerca online :-D)
Se comprendiamo la natura di questi output e li riconosciamo come espressione della volontà
di chi programma statisticamente e amministra[2] il LLM, gli scopi di queste aziende diventano ovvi:
- diffondere false credenze relative a questi software [3]
- ottenere fondi sulla base di queste false credenze (dal mercato, dagli stati etc)
- ottenere normative favorevoli (come l'AI Act) sulla base di queste false credenze
Si tratta di truffatori su scala globale che hanno automatizzato ed affittano la produzione
di "bugie industriali".
Passare da "hallucination" a "bullshit", non riduce molto l'antropomorfizzazione dei software
in questione, ma continua a nascondere la natura dell'operato di chi li produce.
Non siamo di fronte a macchine che imparano ("machine learning"), ma a multinazionali
potentissime che mentono.
Giacomo
[1] a chiunque comprenda come funzioni la selezoone ("previsione" rotfl!) di un token
a partire dal contesto ed il ruolo che vi assume l'input randomico.
[2] ovvero protegge infornaticamente, aggiorna, paga la corrente per programmarli ed eseguirli etc
[3] false credenze che poi vengono ripetute "a pappagallo" da persone incompetenti
(in materia) che purtroppo vengono viste come autorevoli (per esserlo, forse, in altre discipline).
Qualche giorno fa ho ascoltato allibito un'intervista ad un Professore, presentatosi
come fisico del CERN nonché (ex?) docente dell'Università di Torino, snocciolare una
serie incredibile di stronzate sulle "intelligenze artificiali" che parevano uscite da
un convegno transumanista.
Inferendo da un'esperienza vecchia di 20 o 30 anni delle "reti neurali", finiva per prospettare
il sicuro avvento dell'AGI con cui la società si sarebbe dovuta confrontare a livello etico, superando l'idea stessa di anima (sic!) e
riconoscendo l'uomo come un automa complesso del tutto equivalente agli altri.
Si tratta di un'intervista per un podcast italiano dedicato alla "cyber security" (anche se le
competenze di chi lo conduce vacillano continuamente), dunque non orientato a persone
del tutto ignoranti dal punto di vista informatico, ma iper-specializzate, e dunque "vulnerabili"
a questo tipo di disinformazione propagandistica.
June 20, 2024
Re: [nexa] "ChatGPT is bullshit"
by Maria Chiara Pievatolo
On 19/06/24 18:38, Enrico Nardelli wrote:
> Grazie Daniela per questo riferimento.
> La citazione ricordata da Ezra Klein nell'intervista a Gary Marcus, sul
> fatto che una cosa può essere falsa ma, a parte il resto, non inferiore
> alla cosa reale, mi ha fatto venire in mente l'articolo di Daniel Dennet
> sul fatto che generare finzione attraverso l'IA è un atto immorale come
> produrre denaro falso.
E che però rimarrà assai utile, anche quando la bolla scoppierà, e non
solo per giocare a monopoli(i), come ha fatto notare Smowden a proposito
del recente ingresso nel consiglio di amministrazione di OpenAI del
generale in pensione Paul Nakasone "former National Security
Administration (NSA) Director, who also helmed the military's
cybersecurity-focused Cyber Command unit".
"They've gone full mask off: do not ever trust OpenAI or its products,"
Snowden — emphasis his — wrote in a Friday post to X-formerly-Twitter,
adding that "there's only one reason for appointing" an NSA director "to
your board."
"This is a willful, calculated betrayal of the rights of every person on
earth," he continued. "You've been warned."
https://futurism.com/the-byte/snowden-openai-calculated-betrayal
E buonanotte,
MCP
June 19, 2024
Re: [nexa] [Junk released by Allowed List] Re: "ChatGPT is bullshit"
by Daniela Tafani
Il 19/06/2024 18:38, Enrico Nardelli ha scritto:
> Grazie Daniela per questo riferimento.
> La citazione ricordata da Ezra Klein nell'intervista a Gary Marcus, sul fatto che una cosa può essere falsa ma, a parte il resto, non inferiore alla cosa reale, mi ha fatto venire in mente l'articolo di Daniel Dennet sul fatto che generare finzione attraverso l'IA è un atto immorale come produrre denaro falso. https://www.theatlantic.com/technology/archive/2023/05/problem-counterfeit-…
> Ciao, Enrico
>
Già.
"and should be held liable", scriveva Dennet.
Ciao,
Daniela
June 19, 2024
Re: [nexa] "ChatGPT is bullshit"
by Enrico Nardelli
Grazie Daniela per questo riferimento.
La citazione ricordata da Ezra Klein nell'intervista a Gary Marcus, sul
fatto che una cosa può essere falsa ma, a parte il resto, non inferiore
alla cosa reale, mi ha fatto venire in mente l'articolo di Daniel Dennet
sul fatto che generare finzione attraverso l'IA è un atto immorale come
produrre denaro falso.
https://www.theatlantic.com/technology/archive/2023/05/problem-counterfeit-…
Ciao, Enrico
------ Messaggio originale ------
Da "Daniela Tafani" <daniela.tafani(a)unipi.it>
A "J.C. DE MARTIN" <juancarlos.demartin(a)polito.it>; "Nexa"
<nexa(a)server-nexa.polito.it>
Data 19/06/2024 13:15:55
Oggetto Re: [nexa] "ChatGPT is bullshit"
>Molti altri - nessuno dei quali mi pare sia citato nell'articolo - hanno scritto di chatGPT come di un "bullshit generator", in senso tecnico, con riferimento a Frankfurt.
>
>Ad esempio, Dan McQuillan, https://www.danmcquillan.org/chatgpt.html
>
>Anche Richard Stallman usa spesso l'espressione:
>https://stallman.org/archives/2024-jan-apr.html
>
>Per quanto ne so (ma non ho indagato), il primo è stato Ezra Klein, nel 2023, intervistando Gary Marcus:
>
>ChatGPT and systems like it, what they’re going to do right now is they’re going to drive the cost of producing text and images and code and, soon enough, video and audio to basically zero. It’s all going to look and sound and read very, very convincing. That is what these systems are learning how to do. They are learning how to be convincing. They are learning how to sound and seem human.
>
>But they have no actual idea what they are saying or doing. It is bullshit. And I don’t mean bullshit as slang. I mean it in the classic philosophical definition by Harry Frankfurt. It is content that has no real relationship to the truth.
>
>[...]
>
>EZRA KLEIN: Let’s sit on that word truthful for a minute because it gets to, I think, my motivation in the conversation. I’ve been interested — I’m not an A.I. professional the way you are, but I’ve been interested for a long time. I’ve had Sam on the show, had Brian Christian on the show. And I was surprised by my mix of sort of wonder and revulsion when I started using ChatGPT because it is a very, very cool program. And in many ways, I find that its answers are much better than Google for a lot of what I would ask it.
>
>But I know enough about how it works to know that, as you were saying, truthfulness is not one of the dimensions of it. It’s synthesizing. It’s sort of copying. It’s pastiching. And I was trying to understand why I was so unnerved by it. And it got me thinking, have you ever read this great philosophy paper by Harry Frankfurt called “On Bullshit”?
>
>GARY MARCUS: I know the paper.
>
>EZRA KLEIN: So this is a — welcome to the podcast, everybody — this is a philosophy paper about what is bullshit. And he writes, quote, “The essence of bullshit is not that it is false but that it is phony. In order to appreciate this distinction, one must recognize that a fake or a phony need not be in any respect, apart from authenticity itself, inferior to the real thing. What is not genuine may not also be defective in some other way. It may be, after all, an exact copy. What is wrong with a counterfeit is not what it is like, but how it was made.”
>
>And his point is that what’s different between bullshit and a lie is that a lie knows what the truth is and has had to move in the other direction. He has this great line where he says that people telling the truth and people telling lies are playing the same game but on different teams. But bullshit just has no relationship, really, to the truth.
>
>And what unnerved me a bit about ChatGPT was the sense that we are going to drive the cost of bullshit to zero when we have not driven the cost of truthful or accurate or knowledge advancing information lower at all. And I’m curious how you see that concern.
>
>GARY MARCUS: It’s exactly right. These systems have no conception of truth. Sometimes they land on it and sometimes they don’t, but they’re all fundamentally bullshitting in the sense that they’re just saying stuff that other people have said and trying to maximize the probability of that. It’s just auto complete, and auto complete just gives you bullshit.
>
>And it is a very serious problem. I just wrote an essay called something like “The Jurassic Park Moment for A.I.” And that Jurassic Park moment is exactly that. It’s when the price of bullshit reaches zero and people who want to spread misinformation, either politically or maybe just to make a buck, start doing that so prolifically that we can’t tell the difference anymore in what we see between truth and bullshit.
>
>https://www.nytimes.com/2023/01/06/podcasts/transcript-ezra-klein-interview…
>________________________________________
>Da: nexa <nexa-bounces(a)server-nexa.polito.it> per conto di J.C. DE MARTIN <juancarlos.demartin(a)polito.it>
>Inviato: mercoledì 19 giugno 2024 10:51
>A: Nexa
>Oggetto: [nexa] "ChatGPT is bullshit"
>
>ChatGPT is bullshit
>
>Michael Townsen Hicks1 · James Humphries1 · Joe Slater1
>
>Abstract
>Recently, there has been considerable interest in large language models: machine learning systems which produce human- like text and dialogue. Applications of these systems have been plagued by persistent inaccuracies in their output; these are often called “AI hallucinations”. We argue that these falsehoods, and the overall activity of large language models, is better understood as bullshit in the sense explored by Frankfurt (On Bullshit, Princeton, 2005): the models are in an important way indifferent to the truth of their outputs. We distinguish two ways in which the models can be said to be bullshitters, and argue that they clearly meet at least one of these definitions. We further argue that describing AI misrepresentations as bullshit is both a more useful and more accurate way of predicting and discussing the behaviour of these systems.
>
>https://link.springer.com/article/10.1007/s10676-024-09775-5
>
>Hicks, M.T., Humphries, J. & Slater, J. ChatGPT is bullshit. Ethics Inf Technol 26, 38 (2024). https://doi.org/10.1007/s10676-024-09775-5
>
> * Published 08 June 2024
>
> * DOI https://doi.org/10.1007/s10676-024-09775-5
>
-- EN
https://www.hoepli.it/libro/la-rivoluzione-informatica/9788896069516.html
======================================================
Prof. Enrico Nardelli
Past President di "Informatics Europe"
Direttore del Laboratorio Nazionale "Informatica e Scuola" del CINI
Dipartimento di Matematica - Universit� di Roma "Tor Vergata"
Via della Ricerca Scientifica snc - 00133 Roma
home page: https://www.mat.uniroma2.it/~nardelli
blog: https://link-and-think.blogspot.it/
tel: +39 06 7259.4204 fax: +39 06 7259.4699
mobile: +39 335 590.2331 e-mail: nardelli(a)mat.uniroma2.it
online meeting: https://blue.meet.garr.it/b/enr-y7f-t0q-ont
======================================================
--
June 19, 2024