- nexa - server-nexa.polito.it

It is no longer safe to move our governments and societies to US clouds - Bert Hubert's writings
by Alberto Cammozzo March 2, 2025

March 2, 2025

<https://berthub.eu/articles/posts/you-can-no-longer-base-your-government-an…> The very short version: it is madness to continue transferring the running of European societies and governments to American clouds <https://apnews.com/article/trump-inauguration-tech-billionaires-zuckerberg-…>. Not only is it a terrible idea given the kind of things the “King of America <https://www.nbcnews.com/politics/donald-trump/king-trump-rcna192912>” keeps saying, the legal sophistry used to justify such transfers, like the nonsense letter the Dutch cabinet sent last week <https://berthub.eu/tkconv/document.html?nummer=2025D07118>, has now been invalidated <https://www.lawfaremedia.org/article/trump-s-sacking-of-pclob-members-threa…> by Trump himself. And why are we doing this? Convenience. But it is very scary to make yourself 100% dependent on the goodwill of the American government merely because it is convenient. So let’s not. We used to think that grieving went through five stages <https://en.wikipedia.org/wiki/Five_stages_of_grief>: denial, protest, bargaining, depression, and acceptance. We no longer really believe in that theory, but when it comes to dealing with the US, many Europeans are however currently busily negotiating with reality. Some of my more activist friends are already further along in their thinking, and clearly see that it is ridiculous to move our governments over to the American clouds at this stage. They rightly accused me of fooling myself <https://mastodon.nl/@bert_hubert/114015060901981839> by aiming for ‘at least not moving EVERYTHING to America <https://berthub.eu/articles/posts/communicating-without-musk-and-trump-clou…>’, while accepting that we’d still relinquish the majority of our systems. Acknowledging painful changes is not easy <https://fosstodon.org/@bert_hubert/111402692560132040>. In the 1930s, the Netherlands realized the world was becoming pretty dangerous, and therefore planned to buy weapons <https://en.wikipedia.org/wiki/Design_1047_battlecruiser> and ammunition from Germany (!). However, the ordered weaponry was not delivered on time, or at least not in the way you’d want <https://en.wikipedia.org/wiki/German_bombing_of_Rotterdam>. “Negotiating with reality” is for example the letter three Dutch government ministers sent last week <https://berthub.eu/tkconv/document.html?nummer=2025D07118>. Is it wise to report every applicant to your secret service directly to Google, just to get some statistics? The answer the government sent: even if we do that, we don’t, because “Google cannot see the IP address <https://mastodon.nl/@bert_hubert/114031686141633446>”. This is complete nonsense of course, but it’s the kind of thing you tell yourself (or let others tell you) when you don’t want to face reality (or can’t). And why are we reporting all these job applicants to America? Because it’s /cheaper/, they say. The European alternative statistical software, which is already paid for and actually running on the government job site, isn’t as user-friendly either, they say. Nonsense. With all sorts of magic legal spells like “DPIAs <https://www.autoriteitpersoonsgegevens.nl/en/themes/basic-gdpr/gdpr-in-prac…>” and “DTIAs <https://highberg.com/insights/how-do-you-conduct-a-dtia>,” organizations attempt to justify transferring our data and processes to the US. There is a whole industry that has been aiding and abetting this process for years <https://www.privacycompany.eu/>. People also fool themselves that special keys and “servers in the EU” <https://berthub.eu/articles/posts/servers-in-de-eu-eigen-sleutels-helpt-het/> will get you “a safe space <https://www.ncsc.nl/binaries/ncsc/documenten/publicaties/2022/augustus/16/c…>” within the American cloud. It won’t <https://edri.org/our-work/promises-unkept-the-eu-us-data-privacy-framework-…>. All these efforts are not about whether it is /wise/ to hand over control of our data to the US. It is all about if it is /lawful/. But it won’t be lawful either — the legal basis for sharing personal data with American companies is dead since Donald Trump has neutered the special court that would make such transfers legal <https://noyb.eu/en/us-cloud-soon-illegal-trump-punches-first-hole-eu-us-dat…>. We now have the bizarre situation that anyone with any sense can see that America is no longer a reliable partner <https://www.lemonde.fr/en/france/article/2025/02/19/macron-holds-second-day…>, and that the entire large-scale US business world bows to Trump’s dictatorial will <https://apnews.com/article/trump-inauguration-tech-billionaires-zuckerberg-…>, but we STILL are doing everything we can <https://www.utoday.nl/news/75233/waarom-mag-chatgpt-reclame-maken-bij-surf> to transfer entire governments and most of our own businesses to their clouds. Not only is it scary to have all your data available to US spying, it is also a huge risk for your business/government continuity <https://berthub.eu/articles/posts/publicspaces-digitale-autonomie-7-juni-20…>. From now on, all our business processes can be brought to a halt with the push of a button in the US <https://berthub.eu/articles/posts/digitale-soevereiniteit-rijksoverheid-13-…>. And not only will everything then stop, will we ever get our data back? Or are we being held hostage <https://berthub.eu/articles/ransom-cloud.jpeg>? This is not a theoretical scenario, something like this has already happened <https://berthub.eu/articles/posts/servers-in-de-eu-eigen-sleutels-helpt-het…>. Here and there, some parts of at least the Dutch government are deciding not to /migrate EVERYTHING/ to the US <https://berthub.eu/tkconv/document.html?nummer=2025D06767> (kudos to the government workers who are fighting for this!). But even here, the details of Dutch policy are that our data will only ‘for now <https://berthub.eu/tkconv/document.html?nummer=2025D06768>’ stay on our own servers. Experts are also doubtful whether it’s actually possible with the current “partial cloud” plan to keep the data here /exclusively/. And then we come to the apparent reason why we are putting our head on Trump’s chopping block <https://berthub.eu/articles/posts/de-hele-overheid-naar-de-cloud-dat-is-een…>: “American software is just so easy to use”. Personally, I don’t know many fans of MS Teams, Office, and Outlook. We are, however, very /used/ to these software products. We’ve become quite good at using them. But this brings us to the unbearable conclusion that we are entrusting all our data and business processes to the new King of America <https://www.nbcnews.com/politics/donald-trump/king-trump-rcna192912>… because we can’t be bothered to /get used to/ a different word processor <https://www.libreoffice.org/>, or make an effort to support other software <https://berthub.eu/articles/posts/open-source-by-itself-is-no-alternative-f…>. Written down like this, it seems pretty bizarre, and that’s because it is. It’s also not easy to realize that our relationship with America is over after 80 years <https://www.vpro.nl/buitenhof/lees/in-de-uitzending/2025/26-januari.html>. It took time for me as well to accept that. But it really is over. Stop the nonsense of complicated and long-winded legal arguments about why it would be legal to put our citizens’ data in the US. These arguments are not only naïve—they are also no longer true. The US has dismantled the official privacy framework. The deal is dead <https://fosstodon.org/@bert_hubert/114032472494922878>. So stop surrendering our society to America. Have the courage to use different software <https://opendesk.eu/en/ueber/> and services. Even if it takes some getting used to <https://opendesk.eu/en/>. And maybe, once in a while, European governments should have the guts to actually attempt to buy something else. Innovation doesn’t just happen without the promise of revenues <https://berthub.eu/articles/posts/is-europe-just-not-good-at-innovating/>. Spending just a few percent on alternatives would mean billions of euros. That’s a very small price to pay for a society that isn’t 100% dependent on “MAGA”, Donald Trump and Elon Musk. The world is far too dangerous <https://berthub.eu/articles/posts/cyber-security-pre-war-reality-check/> for that.

3 2

A journey through the hyper-political world of microchips
by Alberto Cammozzo Feb. 28, 2025

Feb. 28, 2025

Neutro a chi? <https://www.theguardian.com/technology/2025/feb/28/inside-the-mind-bending-…> A small town in the Netherlands hosts the only factory that produces the only chip-making machines that generate a type of light found nowhere naturally on Earth: extreme ultraviolet, a light emitted by young stars in outer space. This light, known as EUV, is the only way to make one of the world’s most valuable and important technologies at scale: cutting-edge semiconductor chips. The factory is forbidden from selling its EUV machines to China. Below we explain how the chips are made, why they have become the focus of the US-China trade wars, how Taiwan was drawn into the maelstrom, and what could come next. The answers take us from deep underground to outer space, from the dirtiest places in the world to the cleanest, from the hottest temperatures to the coldest, from man-made structures smaller than a virus to equipment so large it takes three planes to move, and finally, to a state in physics that is two opposites at the same time. How are they made? Chips are made up of layers of thin, flat pieces of silicon – called wafers – that hold electric circuits. These circuits are comprised of billions of switches called transistors. Highly complex, powerful chips containing these networks of transistors are commonly referred to as semiconductors. If you want to make semiconductors, you’ll need $380m. This is the cost of the latest EUV machine from Advanced Semiconductor Materials Lithography (ASML). Shipping is a nightmare: the machine is so large and so delicate that it requires 40 freight containers, three cargo planes and 20 trucks to transport it from the Dutch factory in Veldhoven. All this to create and focus light with wavelengths almost as short as X-rays, with enough energy to penetrate solid objects. Chipmakers strive to meet a prediction called Moore’s law: that capacity – or the number of transistors on a chip – tends to double every two years. If chips are to stay the same size, and ideally get smaller, this means that transistors must be ever finer. ASML’s machine carves the patterns into the silicon wafers that hold the transistors. The finer the patterns, the more computing power you can pack on to a chip. Marc Assinck, a company spokesperson, likens light wavelengths to the thickness of a pen stroke. The more detail you want on a page, the thinner your pen should be. EUV light has a wavelength so narrow that it is invisible to the human eye and passes right through most materials. The light is produced by firing a laser at microscopic balls of tin. The tin evaporates into plasma, and the plasma emits light, which is moved through the lithography machine, hitting specially made mirrors. The light is shone through a “mask”, which is the pattern of one layer of a chip, on to the wafer. The area exposed to light hardens and the area not exposed is dissolved in a chemical solution, leaving behind a 3D pattern. Think of a chip like a building with 100 floors. Each building takes four months to produce, and each floor has its own layout, only the features of this layout can be just 25 nanometres: smaller than particles of influenza viruses, which are about 100 nanometres. EUV and other lithography machines carve the patterns of these layers, one by one. The machines aren’t easy to make. Like chips themselves, they’re assembled in dust-free rooms, the cleanest spaces on earth. Chips function at the level of atoms: a single speck of dust can render them useless. Why is Taiwan so important? ASML makes the machines that make chips, but it doesn’t make the chips themselves. That is done, chiefly, by another remarkable company with another unremarkable name: TSMC, or Taiwan Semiconductor Manufacturing Company, which produces nine in every 10 of the world’s most advanced semiconductors – including those that power iPhones. Because Taiwan makes and supplies so many of the US’s semiconductor chips, the US has an added interest in protecting Taiwan amid concerns that China will invade. In 2022, the US convinced the Dutch government to place export controls on ASML’s machines, restricting their sale to China. To date, ASML says, no EUV lithography machines have been shipped to China, which means that unless – or until – it invents its own EUV lithography machines, China will be working with technology a few years older, and less powerful, than that of western countries – deep ultraviolet lithography, for example, instead of extreme ultraviolet. These machines can still produce very complex chips at scale – just not as complex. Artificial intelligence, another technology in which the US and China are fiercely competing to advance, relies on among the world’s most complex and powerful semiconductor chips. The leading designer of these chips is an American company called Nvidia. Its chips are produced by TSMC on machines made by ASML. China’s lack of access to EUV lithography explains why the debut of the Chinese chatbot DeepSeek came as such a shock to markets. A Chinese company produced a product as powerful as Chat GPT with less advanced – cheaper – technology. DeepSeek claims that it cost just $6m to train, compared with the billions of dollars spent by US companies to do the same thing. “The US believes that AI will be a transformative technology, impacting nearly every sector of the economy,” says Chris Miller, the author of Chip War: the Fight for the World’s Most Critical Technology. “So it doesn’t want China to gain an advantage.” It is also crucial for defence and intelligence. The Chinese People’s Liberation Army has made “significant progress” in its efforts to use AI in combat in recent years, according to the Centre for Security and Emerging Technology. But not everyone believes that China’s access to ASML’s machines should be restricted, including ASML. Asked at a Bloomberg conference in October how much the restrictions were really about security threats, its CEO, Christophe Fouquet, said: “More and more people are asking themselves this exact question … Is it really about national security?” The debate might not need to continue for long: in 2024, Chinese company Shanghai Micro Electronics Company (SMIC) revealed that, a year earlier, it had filed a patent for an EUV lithography machine. Where do rare earths come in? China boasts other advantages over western countries in the race to produce chips. In addition to silicon, semiconductors require so-called “rare earths” – in particular, germanium and gallium. By 2030, Gallium demand is projected to increase more than 350% from 2015 levels. Germanium demand is expected to double over the same period. China produces 98% of the world’s raw gallium, and more than two-thirds of the world’s raw germanium. This is one reason why Donald Trump is pressuring Ukraine into handing over its rare earths in exchange for aid, and why, after Trump’s first meeting with Indian prime minister Narendra Modi, the pair announced they had agreed to launch “a recovery and processing initiative” for rare earths. Will quantum chips change everything? Then there are quantum chips. Quantum chips would, in theory, allow computers to solve problems much, much faster than the world’s current super computers. This is because instead of being the equivalent of on or off, or a zero or a one, quantum chips can be both states at once – and every state between. A common explanation is a maze: a normal computer would find the path through a maze by testing each option, one after the other. A quantum computer could test all of them at once. So far, quantum computing has been achieved only in limited circumstances. But Microsoft announced this month that it had built a chip that could mean quantum computers might be built within years rather than decades. Meanwhile, China’s public spending on quantum technology is four times that of the US, according to the Mercator Institute of China Studies, a European thinktank. And the chips are not made with EUV machines. Instead, quantum chips are made by machines that carve patterns into chips using electrons. And China has these machines. China also has a resource often overlooked in the chip debates, says David Reilly, professor of physics and the head of the University of Sydney’s quantum programme. “The key to all of this is people,” he says. Breakthroughs happen because people see a need, and know what the existing ways to meet that need are, and can imagine what they might be. “There are a lot of smart people in China. They produce a lot of Stem graduates,” he says. And those graduates tend to do undergraduate or postgraduate degrees at the top US, Australian, European universities before returning. “I don’t want to say governments are sort of blind to that, but we do focus a lot on the transfer of tangible stuff,” he says. “Inventions don’t happen in a vacuum.”

1 0

Incontro "Anticipatory compliance: come utilizzare i futuri per navigare l'incertezza normativa nell'epoca dei dati" | 21 marzo 2025
by Nexa - Media Feb. 28, 2025

Feb. 28, 2025

Gentilissime, gentilissimi, Vi invitiamo a partecipare ad un evento dal titolo “Anticipatory compliance: come utilizzare i futuri per navigare l'incertezza normativa nell'epoca dei dati”, che si terrà venerdì 21 marzo, dalle ore 14.30 alle ore 17.30. Come può il foresight trasformare l’approccio alla compliance normativa in materia di protezione dei dati personali? Durante questo incontro, attraverso l’analisi di iniziative pionieristiche, verrà esplorato il valore della previsione strategica per affrontare le sfide di un contesto tecnologico e regolamentare incerto ed in continua evoluzione. Interverranno all’incontro: Stefano Leucci (Director – Mobilisights & Fellow – Privacy 4 Futures) Gabriele Rizzo (Executive Partner – Longviews & Fellow – Privacy 4 Futures) Alberto Robiati (Director – ForwardTo) Alessandro Ortalda (Founder – Experirē strategy&advisory & Fellow – Privacy 4 Futures) Giuseppe Vaciago (Partner – 42 Law Firm & Fellow – Privacy 4 Futures) Anna Cataleta (Senior Partner – Partners4Innovation & Fellow – Privacy 4 Futures) Antonio Santangelo (Senior research fellow del Centro Nexa – Università di Torino) L’evento si terrà IN PRESENZA e ONLINE. SEDE FISICA: Centro Nexa su Internet & Società, Politecnico di Torino (Via Boggio 65/a, Torino – 1° piano). Per accedere alla sala si raccomanda di suonare al citofono Portineria e di seguire le indicazioni segnalate lungo il percorso. QUI<https://nexa.polito.it/contatti> maggiori informazioni su come raggiungerci. STANZA VIRTUALE: https://didattica.polito.it/VClass/NexaEvent La partecipazione è libera e gratuita previa registrazione (entro sabato 8 marzo) al seguente link: EVENTBRITE<https://www.eventbrite.com/e/anticipatory-compliance-come-utilizzare-i-futu…> Per maggiori informazioni è possibile consultare la pagina web dedicata all’incontro: https://nexa.polito.it/anticipatory-compliance Vi ringraziamo per l'attenzione, Cordiali saluti, -- Valeria Bergantino Communication Manager Nexa Center for Internet & Society Politecnico di Torino - DAUIN Via Pier Carlo Boggio, 65/A - 10138 Torino web: https://nexa.polito.it/ mail: valeria.bergantino(a)polito.it<mailto:valeria.bergantino@polito.it> tel: 3473443585

1 0

[OT] The Geopolitics of Peace (this is Karmageddon).
by 380° Feb. 28, 2025

Feb. 28, 2025

Buongiorno, scusate per l'OT ma quanto raccontato (sotto) ha avuto un impatto determinante (anche) sul _perché_ e percome la tecnologia - incluso ovviamente "il digitale" - è stata progettata, sviluppata e utilizzata negli ultimi 30 anni. Lo scorso 19 Febbraio Jeffrey Sachs [1] ha tenuto un discorso a mio avviso storico al parlamento europeo, nel quale riassume in modo estremamente efficiente ed efficace cosa è davvero successo dal 1990 fino al 20 Gennaio 2025. Sotto trovate il link e le cose che a mio avviso sono le più significative. Per come lo leggo io, sostanzialmente il suo discorso illustra la (nota?) strategia adottata dagli states per dominare unilateralmente la Terra dopo lo scioglimento del Patto di Varsavia, strategia che è la stessa adottata dalla classe dominante negli ultimi 5000 anni, perché è il suo karma: siccome le risorse sono scarse per sopravvivere dobbiamo (cioè _dovete_) combattere per il Lebensraum (si veda l'ultimo paragrafo della trascrizione sotto). ...ora però è probabile che sia iniziato il Karmageddon! :-D «Welcome to the chaos of the times If you go left and I go right Pray we make it out alive This is Karmageddon» (https://www.youtube.com/watch?v=y29kmnhjtc8) «Professor Jeffrey Sachs: Speech at European Parliament on February 19, 2025» https://www.other-news.info/edited-transcript-professor-jeffrey-sachs-the-g… --8<---------------cut here---------------start------------->8--- Edited transcript of Professor Jeffrey Sachs’ speech in the European Parliament at an event titled “The Geopolitics of Peace”, hosted by former UN Assistant Secretary General and current BSW MEP Michael von der Schulenburg, on February 19, 2025. The transcript has been edited for clarity and annotated in footnotes and hyperlinks. The unedited transcript and audio version is here [2] [...] The United States came to the view, especially during 1990-91, and then with the end of the Soviet Union, that the US now runs the world, and that the US does not have to heed anybody’s views, red-lines, concerns, security viewpoints, international obligations, or any UN framework. I’m sorry to put it so plainly, but I do want you to understand. [...] What happened after 1991, and to bring us to 2008, is that the United States decided that unipolarity meant that NATO would enlarge somewhere from Brussels to Vladivostok, step by step. [...] Neutrality is perhaps the dirtiest word according to the US mindset. If you’re an enemy, we know you’re an enemy. If you are neutral, you are a subversive, because you’re really against us, but just not telling us. You’re only pretending to be neutral. [...] So, the decision was taken by Clinton in 1994 to expand NATO all the way to Ukraine. This is a long-term US project. This is not due to one administration or another. This is a US government project that started more than 30 years ago. [...] Because one thing about America is we always “know” what our counterparts are going to do, and we always get it wrong! And one reason we always get it wrong is that in the non-cooperative game theory that the American strategists play, you don’t actually talk to the other side. [...] A thirty-year project. Ukraine and Georgia were the keys to the project. Why? Because America learned everything it knows from the British. We are the wannabe British Empire. [...] The American political system is a system of image. It’s a system of media manipulation every day. It is a PR system. You could have a president that basically doesn’t function and have that person in power for two years and run for reelection. The one thing is he had to stand on a stage for 90 minutes by himself, and that was the end of it. [...] Bombing Belgrade 78 straight days in 1999 was part of this project. Splitting apart that country when borders are “sacrosanct,” aren’t they? Except for Kosovo, that is. Borders are sacrosanct except when America changes them. Breaking apart Sudan was another related US project. Consider the South Sudan rebellion. Did that just happen because South Sudanese rebelled? Or shall I give you the CIA playbook? Let us please understand as grown-ups what this is about. Military campaigns are costly. They require equipment, training, base camps, intelligence, finance. That support comes from big powers. It doesn’t come from local insurrections. South Sudan did not defeat Sudan in a tribal battle. [...] NATO enlargement, as you know, started in 1999 with Hungary, Poland, and the Czech Republic. Russia was extremely unhappy about it [...] So, these are long-term US projects. It’s wrong to ask, “Is it Clinton? Is it Bush? Is it Obama?” That’s the boring way to look at American politics, as a day-to-day or year-to-year game. Yet that’s not what American politics is. After 1999, the next round of NATO enlargement came in 2004 with seven more countries: the three Baltic states, Romania, Bulgaria, Slovenia, and Slovakia. At this point, Russia was pretty upset. [...] The idea that Putin is reconstructing the Russian empire is childish propaganda. [...] There have been around one hundred regime-change operations by the US since 1947, many in your countries [speaking to the MEPs] and many all over the world.(*16) That’s what the CIA does for a living. Please know it. It’s a very unusual kind of foreign policy. In the American Government, if you don’t like the other side, you don’t negotiate with them, you try to overthrow them, preferably, covertly. If it doesn’t work covertly, you do it overtly. You always say it’s not our fault. They’re the aggressor. They’re the other side. They’re “Hitler.” That comes up every two or three years. Whether it’s Saddam Hussein, whether it’s Assad, whether it’s Putin, that’s very convenient. That’s the only foreign policy explanation the American people are ever given. [...] The mass media repeats it entirely because it’s completely suborned by the US government. [...] [The Maidan Revolution] Ladies and gentlemen, please, how did all those Ukrainian media outlets suddenly appear at the time of the Maidan? Where did all this organization come from? Where did all these buses come from? Where did all those people come from? Are you kidding? This is an organized effort. And it’s not a secret, except perhaps to citizens of Europe and the United States. Everyone else understands it quite clearly. [...] if we put our minds, our resources and our energies towards it, we can transform the world energy system for climate safety. We can protect biodiversity. We can ensure every child gets a quality education. We can do so many wonderful things right now. What do we need for success? In my view, most importantly, we need peace. And my basic point is there are no deep reasons for conflict anywhere because every conflict I study is just a mistake. We are not struggling for Lebensraum. That idea, which essentially came from Malthus and later became a Nazi idea, was always wrong, a fundamental intellectual mistake. We have had race wars, national wars of survival, out of the fear that we don’t have enough for everybody on this planet, so that we are in a struggle for survival. As an economist, I can tell you, we have plenty on the planet for everybody’s sustainable development. Plenty. --8<---------------cut here---------------end--------------->8--- [1] che per i distratti è uno che a _tutti_ i più potenti personaggi geopolicici gli da del tu _e_ che è stato personalmente presente a quasi tutti gli incontri che contano dal 1990 a oggi [2] https://singjupost.com/transcript-jeffrey-sachs-on-the-geopolitics-of-peace… -- 380° (lost in /traslation/) «Welcome to the chaos of the times If you go left and I go right Pray we make it out alive This is Karmageddon»

2 1

11 marzo 2025 - H 09:00/13:00 - Convegno CNR "Le nuove regole deontologiche per i trattamenti ai fini statistici o di ricerca"
by Gianluca Fasano Feb. 28, 2025

Feb. 28, 2025

Gentilissime, gentilissimi, Vi ricordiamo che *martedì 11 marzo*, alle ore 09.00, si terrà il convegno “/Le nuove regole deontologiche per i trattamenti ai fini statistici o di ricerca: sfide e prospettive alla luce dell'evoluzione normativa europea/”, presso la sede del *Consiglio Nazionale delle Ricerche*, piazzale Aldo Moro 7, Roma. L’evento rappresenta un’importante occasione per favorire un dibattito consapevole su una disciplina essenziale per la competitività e l’attrattività della ricerca italiana. Per la prima volta, tali regole saranno definite attraverso un processo di*co-regolazione partecipata*, che coinvolge non solo gli stakeholders e la comunità scientifica, ma anche la società civile. Il convegno è patrocinato dal Garante privacy e dal BBMRI (Nodo Nazionale della Infrastruttura di Ricerca Europea delle Biobanche). Il*programma* dettagliato dell’evento è in allegato. Per confermare la vostra partecipazione, vi preghiamo di registrarvi al seguenteLink registrazione <https://forms.office.com/e/1ct78f4wiX>. Un cordiale saluto. Per il comitato. Gianluca Fasano

1 0

Re: [nexa] Tutto quello che i linguisti computazionali avrebbero voluto sapere sul linguaggio ma non hanno avuto il coraggio di chiedere
by Giuseppe Attardi Feb. 27, 2025

Feb. 27, 2025

Concordo pienamente. Forse manca un riferimento al meccanismo dell’attention, la parte cruciale dei Transformer (Attention is all you need), oltre ai word embedding al loro primo livello, che forniscono la rappresentazione frattale del significato delle parole. La attention è quella che esprime relazioni tra le parole, come quelle a lungo raggio che citi. Sono relazioni sia di tipo sintattico che semantico (anafora) e contestuali. Ad esempio con le sonde sintattiche, Hewitt e Manning mostrano che dalle matrici di attention del livello 6 di un Transformer si può ricostruire l’albero sintattico a dipendenze. Quindi i LLM acquisiscono conoscenze linguistiche, diverse da quelle proposte dai linguisti e in modo diverso da come le insegnamo noi ai bambini e tentavamo di fare coi programmi. Ho seguito discussioni infinite tra linguisti su come scegliere POS e Dependency tags e come annotare questo e quel fenomeno, che alla fine sono risultate del tutto inutili, perché i LLM ne fanno del tutto a meno. Stiamo uscendo dalla fase dell’imitazione da parte delle macchine, quando si cercava di insegnare a loro di imitarci. Ora siamo nella fase della scoperta autonoma, dove le macchine imparano da sole, a modo loro, cose che non conosciamo e che quindi non potremmo insegnargli. — > On 26 Feb 2025, at 12:00, nexa-request(a)server-nexa.polito.it wrote: > > From: Guido Boella <guido.boella(a)unito.it> > To: nexa(a)server-nexa.polito.it > Subject: [nexa] Tutto quello che i linguisti computazionali avrebbero > voluto sapere sul linguaggio ma non hanno avuto il coraggio di > chiedere > Message-ID: > <CAA8c75V1gv4z7Uuoe2qVBLUyM0BoC0YSK=TnNqhu3Y6ba=m1_Q(a)mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > Carissime e carissimi > > segnalo articolo su TuttoScienze > https://www.lastampa.it/tuttoscienze/2025/02/26/news/tutto_quello_che_i_lin… > > la cui versione completa potete trovare sul Magazine Intelligenza > Artificiale magia.news > https://magia.news/tutto-quello-che-i-linguisti-computazionali-avrebbero-vo… > > saluti > Guido

1 0

Consultazione pubblica su "Linee guida per la digitalizzazione 3D di beni storico-artistici e museali" - ParteciPa
by Antonio Vetro' Feb. 27, 2025

Feb. 27, 2025

Segnalo questa consultazione pubblica, qualora in lista ci sia qualcuno che segue la digitalizzazione dei beni artistici e museali: L’obiettivo della consultazione è di raccogliere contributi e commenti sulle Linee guida pe la digitalizzazione 3D di beni storico-artistici e museali che costituiscono un'appendice tecnica delle Linee guida per la Digitalizzazione del patrimonio culturale allegato 1 del Piano nazionale di digitalizzazione (PND) https://partecipa.gov.it/processes/Linee-Guida-Digitalizzazione-3D A.

1 0

Tutto quello che i linguisti computazionali avrebbero voluto sapere sul linguaggio ma non hanno avuto il coraggio di chiedere
by Guido Boella Feb. 26, 2025

Feb. 26, 2025

Carissime e carissimi segnalo articolo su TuttoScienze https://www.lastampa.it/tuttoscienze/2025/02/26/news/tutto_quello_che_i_lin… la cui versione completa potete trovare sul Magazine Intelligenza Artificiale magia.news https://magia.news/tutto-quello-che-i-linguisti-computazionali-avrebbero-vo… saluti Guido -- Dipartimento di Informatica Università di Torino Cso Svizzera 185 10149 Torino Vice-Rettore Vicario dell'Università di Torino per le tematiche dell'Intelligenza Artificiale e 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.chedih.eu www.pai-edih.eu www.firstlife.org www.commonshood.eu https://corpusproject.eu/ www.projectco3.eu www.sipeia.it www.ai-aware.eu nlab4cit.eu [image: CHEDIH] <http://chedih.eu>[image: edih PAI]

3 3

Perché Richard Stallman sbaglia in tema di intelligenza artificiale
by Giuseppe Attardi Feb. 25, 2025

Feb. 25, 2025

LOL, avevo previsto l’obiezione, ma non ti sei accorto della nota: “Potrebbe darsi che ChatGPT abbia letto da qualche parte la spiegazione di ARM.” Comunque anche la spiegazione su Google è vaga, non entra nell’analisi sintattica dettagliata della frase e la risoluzione dell’anafora (la per lei) per spiegarne l’ambiguità, che era il punto del mio messaggio. -- > On 25 Feb 2025, at 12:19, nexa-request(a)server-nexa.polito.it wrote: > Date: Tue, 25 Feb 2025 12:19:26 +0100 > From: Alberto Cammozzo <ac+nexa(a)zeromx.net> > To: nexa(a)server-nexa.polito.it > Subject: Re: [nexa] Perché Richard Stallman sbaglia in tema di > intelligenza artificiale > Message-ID: <7eef1c2c-1947-495a-830b-8aca3b0c0140(a)zeromx.net> > Content-Type: text/plain; charset="utf-8"; Format="flowed" > > "lei suona il violino e lui la tromba" su Google (con le virgolette) mi > da due pagine di risultati, tra cui testi come: > > Si tratta di una tipica anfibologia, ovvero una frase ambigua dal > punto di vista semantico. Altri esempi: “Ho visto mangiare una > gallina” in cui non è chiaro se la gallina mangia o viene mangiata, > oppure il più noto: “Moglie e marito si esibiscono su un palco: lei > suona il violino e lui la tromba”. > > Occorrerebbe provare con testi ambigui che non siano già oggetto di > "spiegazione". > > Alberto > > > On 25/02/25 05:52, Giuseppe Attardi wrote: >> Per anni ho pensato che per capire il linguaggio si dovesse passare >> attraverso l’analisi sintattica e poi per quella semantica, perché >> così dicevano i testi di linguistica. >> Ho sviluppato un parser a dipendenze multiligua basato su word >> embedding e rete neurale che funzionava molto bene. >> L’ultima incarnazione si chiama DiaParser e si trova su GitHub: >> https://github.com/Unipisa/diaparser >> >> Tuttavia John Hewitt ha scoperto, usando un probe sintattico, che le >> matrici di attention dei transformer contengono una rappresentazione >> dell’albero sintattico della frase. >> Altri layer delle metrici corrispondono ad altri fenomeni, come l’anafora. >> I word embedding, che sono al primo livello del transformer, danno una >> rappresentazione molto articolata del significato delle parole, >> inclusi sinonimi e relazioni semantiche tra parole. >> >> A questo punto l’ipotesi che si debba insegnare esplicitamente la >> grammatica e la semantica a un modello ha vacillato. >> Del resto i bambini imparano a capire e parlare una lingua ben prima >> di andare a scuola e che gli vengano insegnati i concetti linguistici >> astratti. >> Probabilmente l’esposizione a 30 milioni di frasi nei primi tre anni >> di vita è sufficiente (c’è chi dice siano di più, tenendo conto anche >> dei feedback visivi). >> >> Quindi, mai dare per scontate certe ipotesi che ci sembrano ovvie. >> >> Ho provato quindi a chiedere questo a ChatGPT. >> >> >> Analizza questa frase: lei suona il violino e lui la tromba. Cosa fa >> lui esattamente? >> >> >> ChatGPT: La frase è ambigua perché la costruzione grammaticale lascia >> spazio a due interpretazioni: >> >> 1. Interpretazione comune e probabile: “Lei suona il violino e lui >> [suona] la tromba.” Qui si sottintende il verbo “suonare” anche per il >> secondo soggetto, quindi si capisce che entrambi stanno suonando >> strumenti musicali. >> >> 2. Interpretazione ambigua e comica: “Lei suona il violino e lui la >> tromba” potrebbe essere letto in modo diverso, dove “la tromba” non è >> lo strumento musicale ma un’azione riferita a “lei”. In italiano >> colloquiale, “trombare” può avere un significato volgare (sinonimo di >> “fare sesso”). Se si legge in questo modo, la frase assume un doppio >> senso ironico. >> >> >> Il contesto in cui viene detta o scritta la frase aiuta a chiarire >> quale delle due interpretazioni sia corretta. >> >> >> Potrebbe darsi che ChatGPT abbia letto da qualche parte la spiegazione >> di ARM. >> >> Ma anche concludere che il modello non sa fare certe cose potrebbe >> essere azzardato. >> >> >>> On 14 Feb 2025, at 16:59, Angelo Raffaele Meo <angelo.meo(a)polito.it> >>> wrote: >>> >>> >>> Difendo Richard Stallman ed anche l'Intelligenza Artificiale. >>> Nel 1960, quando quasi tutti i partecipanti a questo dibattito non >>> erano ancora nati oppure frequentavano l'asilo, io realizzai uno dei >>> primi prototipi di quel capitolo dell'Intelligenza Artificiale che >>> chiamiamo "riconoscimento della voce". Riconosceva sotto dettatura >>> una ventina di parole distinte, isolate, basandosi sull'analisi >>> acustico-lessicale del segnale vocale. Ad esempio, i brevissimi >>> istanti di silenzio caratteristici della lettera esplosiva "t" >>> suggerivano la parola "sette" oppure la parola "otto", mentre le >>> vocali hanno spettri di frequenza diversi. >>> In quella prima fase della ricerca comprendemmo, ad esempio, che la >>> dettatura automatica o la traduzione da una lingua ad un'altra >>> richiedono l'integrazione dell'analisi lessicale con l'analisi >>> sintattica, ma è ancora più importante l'analisi semantica. >>> Fui io ad introdurre un esempio significativo che è stato citato più >>> volte. Consideriamo la frase "lei suona il piano e lui la tromba". >>> Due diversi alberi sintattici possono sottendere quella frase. Nel >>> primo albero sintattico la parola "tromba" è un'unità nominale; nel >>> secondo albero sintattico la stessa parola è unità verbale. >>> Per l'ascoltatore che non sa nulla della vita sentimentale di lei e >>> lui, "tromba" è un'unità nominale; invece, per l'ascoltatore che sa >>> che lei e lui sono amanti, in un particolare contesto della >>> conversazione, in virtù di quel capitolo della semantica che è >>> chiamato "pragmatica", "tromba" diventa un'unità verbale, con >>> implicazioni importanti sul processo di traduzione. >>> In questo momento, introdurre la semantica o la pragmatica >>> nell'elaborazione del linguaggio naturale è ancora difficilissimo. >>> Quindi, Intelligenza Artificiale interpretata alla lettera non esiste >>> ancora. Sono sicuro che questo fosse il significato della >>> provocatoria affermazione di Stallman e quindi che Stallman abbia >>> ragione. >>> Comunque, senza usare semantica o pragmatica, in virtù di una >>> raffinata elaborazione statistica e della capacità tecnologica di >>> trattare immensi volumi di dati, quel nuovo insieme di discipline >>> scientifiche che chiamiamo Intelligenza Artificiale sta producendo >>> risultati meravigliosi. La dettatura automatica e la traduzione da >>> una lingua all'altra sono due esempi significativi. Commettono ancora >>> qualche errorino ma sono di grandissima utilità. >>> Viva l'Intelligenza Artificiale, chiara dimostrazione >>> dell'intelligenza naturale di homo sapiens. >>> ------------------------------------------------------------------------ >>> *From:* nexa <nexa-bounces(a)server-nexa.polito.it> on behalf of >>> Giuseppe Attardi <attardi(a)di.unipi.it> >>> *Sent:* Friday, February 14, 2025 12:20 PM >>> *To:* nexa(a)server-nexa.polito.it <nexa(a)server-nexa.polito.it> >>> *Subject:* Re: [nexa] Perché Richard Stallman sbaglia in tema di >>> intelligenza artificiale >>> È anche peggio di così, come ammettono nei manuali per medici: >>> >>> I medici spesso utilizzano termini vaghi come "altamente probabile", >>> "improbabile" o "non è da escludere" per descrivere la possibilità di >>> una malattia. >>> >>> <default_share_msd.jpg> >>> Strategie per il processo decisionale clinico - Argomenti speciali - >>> Manuali MSD Edizione Professionisti >>> <https://www.msdmanuals.com/it/professionale/argomenti-speciali/processo-dec…> >>> msdmanuals.com >>> <https://www.msdmanuals.com/it/professionale/argomenti-speciali/processo-dec…> >>> >>> <https://www.msdmanuals.com/it/professionale/argomenti-speciali/processo-dec…> >>> >>> >>>> On 14 Feb 2025, at 12:00, nexa-request(a)server-nexa.polito.it wrote: >>>> >>>> From: Giacomo Tesio <giacomo(a)tesio.it> >>>> To: nexa(a)server-nexa.polito.it >>>> Subject: Re: [nexa] Perché Richard Stallman sbaglia in tema di >>>> intelligenza artificiale >>>> Message-ID: <5FABC401-5B15-4A28-9850-5BC9ED4E2ED6(a)tesio.it> >>>> Content-Type: text/plain; charset=UTF-8 >>>> >>>> Giuseppe, mi permetto un buon consiglio non richiesto: >>>> >>>> Il 14 Febbraio 2025 10:36:55 UTC, Giuseppe Attardi ha scritto: >>>> >>>>> Quante probabilità ho che sia questo o quello? >>>>> >>>>> Quant’è la probabilità di guarigione? >>>>> >>>>> Quanto stima possa durare la riabilitazione? >>>>> >>>>> Quant’è il rischio di ricadute? >>>>> >>>>> >>>>> È tutto un ragionare a spanne su probabilità stimate a naso. >>>> >>>> se il tuo medico risponde a domande come queste con stime numeriche >>>> delle probabilità (a naso o meno), cambia medico. >

2 1

Re: [nexa] Perché Richard Stallman sbaglia in tema di intelligenza artificiale
by alessandro marzocchi Feb. 25, 2025

Feb. 25, 2025

I bambini come imparano a capire e parlare una lingua? Domanda Attardi e questo, a mio avviso, è il punto da cui partire: noi umani siamo così diversi da queste macchine? I bambini come identificano la persona materna, perchè chiamano "mamma" questa persona? Pare avvenga in molte. Mamma potrebbe essere il nome di un qualunque oggetto, invece perchè "mamma" è il nome della mamma? In questo processo, la "statistica" non rileva? I bambini non imparano le parole con tentativi continui, con continui premi e correzioni? Fino ad arrivare ai grandi della scienza la verità non si scopre provando e riprovando? ----- Date: Tue, 25 Feb 2025 05:52:42 +0100 From: Giuseppe Attardi <attardi(a)di.unipi.it> An HTML attachment was scrubbed... URL: < http://server-nexa.polito.it/pipermail/nexa/attachments/20250225/0832018c/a… > > i bambini imparano a capire e parlare una lingua ben prima > di andare a scuola e che gli vengano insegnati i concetti > linguistici astratti ----- Le altre questioni che ritengo rilevanti sono: - una macchina potrà indagare con mezzi suoi propri? - rispettando il primo principio di Asimov? Cordialmente. Duccio (Alessandro Marzocchi)

1 0