- nexa - server-nexa.polito.it

ChatGPT... dono degli dei? (was: ChatGPT disabled for users in Italy)
by Giacomo Tesio April 10, 2023

April 10, 2023

Salve Anna, grazie della condivisione. Ho avuto modo di ascoltare Carlo con attenzione tramite Piped https://piped.garudalinux.org/watch?v=phcGTeIOwjY Sono un umile programmatore e sono certo che i filosofi in lista mi potranno correggere, ma non posso fare a meno di notare un difetto fondamentale nel parallelo proposto da Carlo fra il dono della scrittura agli egiziani e ChatGPT: Open AI non è un dio. Open AI è una società sostanzialmente di Microsoft. Ora, lungi da me voler mettere in discussione la fede religiosa di Carlo, quale che essa sia, ma come informatico non ho potuto non notare una profonda confusione (Carlo ad un certo punto la chiama "offuscamento") su ciò di cui stiamo parlando. """ la risposta di Chat GPT, che tra l'altro è una risposta uno a uno, cioé non è che Chat GPT pubblica sul web, cioé non è un motore di ricerca. Poi noi lo usiamo come un motore di ricerca, ma è una conversazione uno a uno di un umano con una macchina... """ Open AI pubblica Chat GPT sul Web. Il fatto che non sia un motore di ricerca è irrilevante: anche https://labtf.rime.inrim.it/webClock_utcit.html non è un motore di ricerca, ma INRiM pubblica su quella pagina web l'ora esatta calcolata dai propri orologi atomici. Non è Chat GPT è pubblicare qualcosa: è un software pubblicato sul Web, tant'é che è accessibile attraverso un browser. Anche le API di ChatGPT sono pubblicate via HTTP: https://platform.openai.com/docs/api-reference/authentication Open AI dunque pubblica Chat GPT sul Web. Carlo mostra poi una profonda confusione sulla natura di questo servizio, definendolo una "conversazione uno a uno di un umano con una macchina". Mi sento responsabile di questa confusione, per aver dato per scontato un aspetto fondamentale nello schema che proposi qualche giorno fa. Quando N persone, separatamente, scrivono sul sito di ChatGPT il proprio nome succede questo: Web + Open AI -----------+ | o < Carlo Blengino | | |____.\_ /|\ | o | / | / \ | /|\ | / | | / \ +-----------+ / | | | |---+ | | o | | | o < Brian Hood | /|\ | ChatGPT |--------|____.\_ /|\ | / \ | | | / \ | | |---+ | | o +-----------+ \ | | /|\ | \ | o < Stefano Quintarelli | / \ | \ |____.\_ /|\ | | | / \ +--------------------+ | | | Ciascuno invia al sito web di Open AI chiamato ChatGPT che risponde a ciascuno con una diversa storia statisticamente correlata (non "inventata") con ciò che hanno fatto nella vita. Ma sono le persone di Open AI ad aver deciso (statisticamente ovvero un po'... "alla carlona" :-D) cosa Chat GPT risponderà a ciascuno! Per comprendere che funzione svolge ChatGPT propongo un parallelo molto più terra terra rispetto al mito del dio Teuth. Quando video-chiamiamo qualcuno, non vediamo il nostro interlocutore ma una sua rappresentazione. Tale rappresentazione è mediata da innumerevoli automatismi e può essere alterata dai software che la veicolano: ad esempio lo sfondo che vediamo potrebbe essere fittizio o sfumato. Questa alterazione è introdotta da un software al servizio del nostro interlocutore. Con Chat GPT fa lo stesso. ChatGPT non permette di dialogare con GPT4, ma di dialogare con le persone che dirigono Open AI, che hanno creato GPT4 per poter dialogare con milioni di persone in contemporanea. Oltre agli innumerevoli aspetti tecnici, hanno letteralmente scelto i contenuti da utilizzare per la programmazione statistica (che non sono tutti quelli esistenti o disponibili) decidendo, in modo più o meno attento, ciò che quel software avrebbe detto PER LORO. Se gliene diamo il merito, dobbiamo anche attribuirgliene la responsabilità. Ne deriva che ciò che Carlo confonde con una "conversazione uno a uno di un umano con una macchina" è in realtà una conversazione pochi a molti, in cui pochi utilizzano un proxy automatizzato per fare meno fatica, ACCETTANDO CHE SCRIVA AL LORO POSTO. ChatGPT non è diverso dal filtro per lo sfondo, insomma. Carlo deriva da questa confusione sulla cardinalità della conversazione conclusioni giuridiche piuttosto bizzarre: """ tutti questi errori avvengono dove? non tanto sul fatto che questo ti dica delle stupidaggini su "se la terra è piatta o tonda" ma sul fatto che gli metti dentro "Carlo Blengino" e questo si inventa una biografia che, se va bene, non centra niente. MA LO DICE A ME, tra l'altro, e quindi se dice che io sono un pedofilo lo dice a me. Poi certo che se io prendo quella roba e dico "AH ChatGPT ha detto..." Ha detto a chi? Ha detto a me! """ Il problema di questa analisi è la prospettiva individuale utilizzata. Che succede se le stesse tre persone chiedono, separatamente, ma prima che venga in qualsivoglia modo alterato il modello di GPT4 la stessa cosa che avevano chiesto i concittadini di Brian Hood? Web + Open AI -----------+ | o < Brian Hood | | |____.\_ /|\ | o | / | / \ | /|\ | / | | / \ +-----------+ / | | | |---+ | | o | | | o < Brian Hood | /|\ | ChatGPT |--------|____.\_ /|\ | / \ | | | / \ | | |---+ | | o +-----------+ \ | | /|\ | \ | o < Brian Hood | / \ | \ |____.\_ /|\ | | | / \ +--------------------+ | | Tutti riceveranno pressappoco la stessa risposta. Dunque Open AI sta diffondendo informazioni false e diffamatorie su alcune persone in questo momento a N altre persone che ne fanno richiesta. E questa è diffamazione. Il fatto che alcune persone si limitino a chiedere informazioni su sé stesse e che le risposte risultino evidentemente false, ha reso il problema evidente, ma NON E' il problema. Ma attenzione: non è ChatGPT (o GPT4) a diffamare in modo subdolo e imprevedibile persone innocenti rispondendo ad allocchi! E' Open AI! Ovvero le persone che amministrano e governano tale software. Sta rispondendo per loro: se spegnessero ChatGPT e rispondessero di persona ci metterebbero più tempo, ed eviterebbero le stronzate statistiche, ma l'interfaccia Web non richiederebbe alcuna modifica. Io non riesco davvero a capacitarmi di come una persona intelligente e preparata come Carlo possa perdere di vista le persone che si nascondono dietro la macchina. E la cosa, ad essere onesti mi preoccupa. Non siamo di fronte ad un'invenzione che possa semplicemente mettere in crisi un sistema oscurantista come il copyright. Siamo di fronte ad uno strumento PROGETTATO per ingannare le persone, individualmente e su larga scala. E l'inganno non sta solo nel "gioco di imitazione" dell'intelligenza umana. L'inganno sta proprio nel nascondere alla coscienza di persone intelligenti e in buona fede, le responsabilità delle persone che realizzano e controllano questi strumenti, perseguendo il proprio interesse. Non abbiamo di fronte pericolose intelligenze artificiali generali o divinità benevole. Abbiamo di fronte criminali che sfruttano la straordinaria ignoranza informatica diffusa per cercare di dominare il mondo. Vi prego, adottate un minimo di spirito critico. Studiate informatica! E quando avrete compreso come funzionano questi sistemi, chiedetevi "cui prodest?" Giacomo

1 0

Re: [nexa] Statement from the listed authors of Stochastic Parrots on the “AI pause” letter (Enrico Nardelli)
by alessandro marzocchi April 10, 2023

April 10, 2023

Regulatory efforts should focus on transparency, accountability and preventing exploitative labor practices./ : è quanto necessario e sufficiente, sulla trasparenza preciserei l'obbligo di programmi interamente aperti. Valutata dalla prospettiva marketing, ci stiamo appassonando per un'operazione da manuale. Cordialmente. Duccio (Alessandro Marzocchi) Date: Mon, 10 Apr 2023 20:00:14 +0200 > From: Enrico Nardelli <nardelli(a)mat.uniroma2.it> > (mi pare non sia passato in lista) > Statement from the listed authors of Stochastic Parrots on the “AI pause” > letter > Timnit Gebru (DAIR), Emily M. Bender (University of Washington), Angelina > McMillan-Major (University of Washington), Margaret Mitchell (Hugging Face) > March 31, 2023 > /Tl;dr: The harms from so-called AI are real and present and follow from > the acts of people and corporations deploying automated systems. Regulatory > efforts should focus on transparency, accountability and preventing > exploitative labor practices./ > https://www.dair-institute.org/blog/letter-statement-March2023/ > >

1 0

Statement from the listed authors of Stochastic Parrots on the “AI pause” letter
by Enrico Nardelli April 10, 2023

April 10, 2023

(mi pare non sia passato in lista) Statement from the listed authors of Stochastic Parrots on the “AI pause” letter Timnit Gebru (DAIR), Emily M. Bender (University of Washington), Angelina McMillan-Major (University of Washington), Margaret Mitchell (Hugging Face) March 31, 2023 /Tl;dr: The harms from so-called AI are real and present and follow from the acts of people and corporations deploying automated systems. Regulatory efforts should focus on transparency, accountability and preventing exploitative labor practices./ https://www.dair-institute.org/blog/letter-statement-March2023/ / -- -- EN https://www.hoepli.it/libro/la-rivoluzione-informatica/9788896069516.html ====================================================== Prof. Enrico Nardelli Presidente di "Informatics Europe" Direttore del Laboratorio Nazionale "Informatica e Scuola" del CINI Dipartimento di Matematica - Università 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 ====================================================== --

1 0

"Pausing AI Developments Isn't Enough. We Need to Shut it All Down"
by J.C. DE MARTIN April 10, 2023

April 10, 2023

*Pausing AI Developments Isn't Enough. We Need to Shut it All Down* BY ELIEZER YUDKOWSKY MARCH 29, 2023 6:01 PM EDT / Yudkowsky is a decision theorist from the U.S. and leads research at the Machine Intelligence Research Institute. He's been working on aligning Artificial General Intelligence since 2001 and is widely regarded as a founder of the field./ An open letter published today calls for “all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4.” This 6-month moratorium would be better than no moratorium. I have respect for everyone who stepped up and signed it. It’s an improvement on the margin. I refrained from signing because I think the letter is understating the seriousness of the situation and asking for too little to solve it. The key issue is not “human-competitive” intelligence (as the open letter puts it); it’s what happens after AI gets to smarter-than-human intelligence. Key thresholds there may not be obvious, we definitely can’t calculate in advance what happens when, and it currently seems imaginable that a research lab would cross critical lines without noticing. continua qui: https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/

5 6

"DPReview.com to close"
by J.C. DE MARTIN April 9, 2023

April 9, 2023

Dopo quasi 25 anni chiude il noto sito di fotografia Dpreview: https://www.dpreview.com/news/5901145460/dpreview-com-to-close L'aspetto che trovo stupefacente, anzi, oltraggioso (tranne che per i fanatici della proprietà privata, ovviamente), è che - a quanto pare - il sito /scomparirà/ proprio. Quindi l'enorme numero di recensioni, iniziate prima della diffusione della fotografia digitale, e l'ancor più grande numero di commenti degli utenti (non di rado con foto), _scomparirà nel nulla_. Come qualcuno ha scritto, è come se scomparisse la (o, almeno, una) "Wikipedia della fotografia". juan carlos

6 13

OpenAI threatened with landmark defamation lawsuit over ChatGPT false claims
by Alberto Cammozzo April 9, 2023

April 9, 2023

<https://arstechnica.com/tech-policy/2023/04/openai-may-be-sued-after-chatgp…> Update, 8:30 pm: A spokesperson for Gordon Legal provided a statement to Ars confirming that responses to text prompts generated by ChatGPT 3.5 and 4 vary, with defamatory comments still currently being generated in ChatGPT 3.5. Among "several false statements" generated by ChatGPT were falsehoods stating that Brian Hood "was accused of bribing officials in Malaysia, Indonesia, and Vietnam between 1999 and 2005, that he was sentenced to 30 months in prison after pleading guilty to two counts of false accounting under the Corporations Act in 2012, and that he authorised payments to a Malaysian arms dealer acting as a middleman to secure a contract with the Malaysian Government." Because "all of these statements are false," Gordon Legal "filed a Concerns Notice to OpenAI" that detailed the inaccuracy and demanded a rectification. “As artificial intelligence becomes increasingly integrated into our society, the accuracy of the information provided by these services will come under close legal scrutiny," James Naughton, Hood's lawyer, said, noting that if a defamation claim is raised, it "will aim to remedy the harm caused" to Hood and "ensure the accuracy of this software in his case.”) It was only a matter of time before ChatGPT—an artificial intelligence tool that generates responses based on user text prompts—was threatened with its first defamation lawsuit. That happened last month, Reuters reported today, when an Australian regional mayor, Brian Hood, sent a letter on March 21 to the tool’s developer, OpenAI, announcing his plan to sue the company for ChatGPT’s alleged role in spreading false claims that he had gone to prison for bribery. To avoid the landmark lawsuit, Hood gave OpenAI 28 days to modify ChatGPT’s responses and stop the tool from spouting disinformation. According to Hood’s legal team, ChatGPT could seriously damage the mayor’s reputation by falsely claiming that Hood had been convicted for taking part in a foreign bribery scandal in the early 2000s while working for a subsidiary of the Reserve Bank of Australia. Hood had worked for a subsidiary, Note Printing Australia, but rather than being found guilty of bribery, Hood was the one who notified authorities about the bribes. Reuters reported that Hood was never charged with any crimes, but ChatGPT seems to have confused the facts when generating some responses to text prompts inquiring about Hood's history. OpenAI did not immediately respond to Ars’ request for comment. Ars attempted to replicate the error using ChatGPT, though, and it seems possible that OpenAI has fixed the errors as Hood's legal team has directed. When Ars asked ChatGPT if Hood served prison time for bribery, ChatGPT responded that Hood “has not served any prison time” and clarified that “there is no information available online to suggest that he has been convicted of any criminal offense.” Ars then asked if Hood had ever been charged with bribery, and ChatGPT responded, “I do not have any information indicating that Brian Hood, the current mayor of Hepburn Shire in Victoria, Australia, has been charged with bribery.” Ars could not immediately reach Hood’s legal team to find out which text prompts generated the alleged defamatory claims or to confirm if OpenAI had responded to confirm that the error had been fixed. The legal team was still waiting for that response at the time that Reuters' report published early this morning. Hood’s lawyer, James Naughton, a partner at Gordon Legal, told Reuters that Hood’s reputation is “central to his role” as an elected official known for “shining a light on corporate misconduct.” If AI tools like ChatGPT threaten to damage that reputation, Naughton told Reuters, “it makes a difference to him." That's why the landmark defamation lawsuit could be his only course of action if the alleged ChatGPT-generated errors are not corrected, he said. It's unclear to Hood how many people using ChatGPT were exposed to the disinformation. Naughton told Reuters that the defamatory statements were so serious that Hood could claim more than $130,000 in defamation damages under Australian law. Whether companies like OpenAI could be held liable for defamation is still debatable. It’s possible that companies could add sufficient disclaimers to products to avoid such liability, and they could then pass the liability on to users, who could be found to be negligently or intentionally spreading false claims while knowing that ChatGPT cannot always be trusted. Australia has recently drawn criticism for how it has reviewed defamation claims in the digital age. In 2020, Australia moved to redraft its defamation laws after a high court ruling found that publishers using social media platforms like Facebook should be held liable for defamatory third-party comments on their pages, CNBC reported in 2021. That is contrary to laws providing immunity shields for platforms, such as Section 230 in the US. At that time, Australia considered the question of whether online publishers should be liable for defamatory statements made by commenters in online forums “one of the most complex to address,” with “complications beyond defamation law alone.” By the end of last year, Australian attorneys general were pushing new reforms to ensure that publishers could avoid any liability, The Guardian reported. Now it looks like new generative AI tools like ChatGPT that publish potentially defamatory content will likely pose the next complex question—one that regulators, who are just now wrapping their heads around publisher liability on social media, may not yet be prepared to address. Naughton told Reuters that if Hood’s lawsuit proceeds, it would accuse OpenAI of “giving users a false sense of accuracy by failing to include footnotes” and failing to inform users how ChatGPT's algorithm works to come up with answers that may not be completely accurate. AI ethics experts have urged regulators to ensure that companies like OpenAI are more transparent about how AI tools work. If OpenAI doesn't adequately respond to Hood's concerns, his lawsuit could proceed before the laws clarify who is responsible for alleged AI-generated defamation. "It would potentially be a landmark moment in the sense that it's applying this defamation law to a new area of artificial intelligence and publication in the IT space," Naughton told Reuters.

2 1

Non c’è crisi tecnologica in Silicon Valley
by de petra giulio April 9, 2023

April 9, 2023

Intervistato e intervistatore entrambi nexiani https://centroriformastato.it/non-ce-crisi-tecnologica-nella-silicon-valley/

6 6

Tesla workers shared ‘intimate’ car camera images, ex-employees allege: ‘Massive invasion of privacy’
by Alberto Cammozzo April 7, 2023

April 7, 2023

<https://www.theguardian.com/technology/2023/apr/07/tesla-intimate-car-camer… Tesla assures its millions of electric car owners that their privacy “is and will always be enormously important to us”. The cameras it builds into vehicles to assist driving, it notes on its website, are “designed from the ground up to protect your privacy”. But between 2019 and 2022, groups of Tesla employees privately shared via an internal messaging system sometimes highly invasive videos and images recorded by customers’ car cameras, according to interviews by Reuters with nine former employees. Some of the recordings caught Tesla customers in embarrassing situations. One ex-employee described a video of a man approaching a vehicle completely naked. Also shared: crashes and road-rage incidents. One crash video in 2021 showed a Tesla driving at high speed in a residential area hitting a child riding a bike, according to another ex-employee. The child flew in one direction, the bike in another. The video spread around a Tesla office in San Mateo, California, via private one-on-one chats, “like wildfire”, the ex-employee said. I saw some scandalous stuff sometimes - scenes of intimacy, a lot of stuff that I wouldn’t want anybody to see about my life Other images were more mundane, such as pictures of dogs and funny road signs that employees made into memes by embellishing them with amusing captions or commentary, before posting them in private group chats. While some postings were only shared between two employees, others could be seen by scores of them, according to several ex-employees. Tesla states in its online customer privacy notice that its “camera recordings remain anonymous and are not linked to you or your vehicle”. But seven former employees told Reuters the computer program they used at work could show the location of recordings, which potentially could reveal where a Tesla owner lived. One ex-employee also said that some recordings appeared to have been made when cars were parked and turned off. Several years ago, Tesla would receive video recordings from its vehicles even when they were off, if owners gave consent. It has since stopped doing so. “We could see inside people’s garages and their private properties,” said another former employee. “Let’s say that a Tesla customer had something in their garage that was distinctive, you know, people would post those kinds of things.” Tesla didn’t respond to detailed questions sent to the company for this report. About three years ago, some employees stumbled upon and shared a video of a unique submersible vehicle parked inside a garage, according to two people who viewed it. Nicknamed Wet Nellie, the white Lotus Esprit sub had been featured in the 1977 James Bond film, The Spy Who Loved Me. The vehicle’s owner: Tesla chief executive Elon Musk, who had bought it for about $968,000 at an auction in 2013. It is not clear whether Musk was aware of the video or that it had been shared. Musk didn’t respond to a request for comment. To report this story, Reuters contacted more than 300 former Tesla employees who had worked at the company over the past nine years and were involved in developing its self-driving system. More than a dozen agreed to answer questions, all speaking on condition of anonymity. Employees passed around images captured by a Tesla camera of a James Bond vehicle owned by Elon Musk. Reuters wasn’t able to obtain any of the shared videos or images, which ex-employees said they hadn’t kept. The news agency also wasn’t able to determine if the practice of sharing recordings, which occurred within some parts of Tesla as recently as last year, continues today or how widespread it was. Some former employees contacted said the only sharing they observed was for legitimate work purposes, such as seeking assistance from colleagues or supervisors. Labeling pedestrians and street signs The sharing of sensitive videos illustrates one of the less-noted features of artificial intelligence systems: they often require armies of human beings to help train machines to learn automated tasks such as driving. Since about 2016, Tesla has employed hundreds of people in Africa and later the United States to label images to help its cars learn how to recognize pedestrians, street signs, construction vehicles, garage doors, and other objects encountered on the road or at customers’ houses. To accomplish that, data labelers were given access to thousands of videos or images recorded by car cameras that they would view and identify objects. Tesla increasingly has been automating the process, and shut down a data-labeling hub last year in San Mateo, California. But it continues to employ hundreds of data labelers in Buffalo, New York. In February, Tesla said the staff there had grown 54% over the previous six months to 675. Two ex-employees said they weren’t bothered by the sharing of images, saying that customers had given their consent or that people long ago had given up any reasonable expectation of keeping personal data private. Three others, however, said they were troubled by it. “It was a breach of privacy, to be honest. And I always joked that I would never buy a Tesla after seeing how they treated some of these people,” said one former employee. Another said: “I’m bothered by it because the people who buy the car, I don’t think they know that their privacy is, like, not respected. We could see them doing laundry and really intimate things. We could see their kids.” One former employee saw nothing wrong with sharing images, but described a function that allowed data labelers to view the location of recordings on Google Maps as a “massive invasion of privacy”. David Choffnes, executive director of the Cybersecurity and Privacy Institute at Northeastern University in Boston, called the sharing of sensitive videos and images by Tesla employees “morally reprehensible”. “Any normal human being would be appalled by this,” he said. He noted that circulating sensitive and personal content could be construed as a violation of Tesla’s own privacy policy, potentially resulting in intervention by the US Federal Trade Commission (FTC), which enforces federal laws relating to consumers’ privacy. A spokesperson for the FTC said it doesn’t comment on individual companies or their conduct. To develop self-driving car technology, Tesla collects a vast trove of data from its global fleet of several million vehicles. The company requires car owners to grant permission on the cars’ touchscreens before Tesla collects their vehicles’ data. “Your Data Belongs to You,” states Tesla’s website. In its customer privacy notice, Tesla explains that if a customer agrees to share data, “your vehicle may collect the data and make it available to Tesla for analysis. This analysis helps Tesla improve its products, features, and diagnose problems quicker.” It also states that the data may include “short video clips or images”, but isn’t linked to a customer’s account or vehicle identification number, “and does not identify you personally”. A camera on a Tesla electric vehicle. A camera on a Tesla electric vehicle. Photograph: Aly Song/Reuters Carlo Piltz, a data privacy lawyer in Germany, told Reuters it would be difficult to find a legal justification under Europe’s data protection and privacy laws for vehicle recordings to be circulated internally when it has “nothing to do with the provision of a safe or secure car or the functionality” of Tesla’s self-driving system. In recent years, Tesla’s car-camera system has drawn controversy. In China, some government compounds and residential neighborhoods have banned Teslas because of concerns about its cameras. In response, Musk said in a virtual talk at a Chinese forum in 2021: “If Tesla used cars to spy in China or anywhere, we will get shut down.” Elsewhere, regulators have scrutinized the Tesla system over potential privacy violations. But the privacy cases have tended to focus not on the rights of Tesla owners but of passersby unaware that they might be being recorded by parked Tesla vehicles. In February, the Dutch Data Protection Authority (DPA) said it had concluded an investigation of Tesla over possible privacy violations regarding “sentry mode”, a feature designed to record any suspicious activity when a car is parked and alert the owner. The watchdog determined it wasn’t Tesla, but the vehicles’ owners, who were legally responsible for their cars’ recordings. It said it decided not to fine the company after Tesla said it had made several changes to sentry mode, including having a vehicle’s headlights pulse to inform passersby that they may be being recorded. A DPA spokesperson declined to comment on Reuters’ findings, but said in an email: “Personal data must be used for a specific purpose, and sensitive personal data must be protected.” Tesla calls its automated driving system “autopilot”. Introduced in 2015, the system included such advanced features as allowing drivers to change lanes by tapping a turn signal and parallel parking on command. To make the system work, Tesla initially installed sonar sensors, radar and a single front-facing camera at the top of the windshield. A subsequent version, introduced in 2016, included eight cameras all around the car to collect more data and offer more capabilities. Musk’s future vision is eventually to offer a “full self-driving mode” that would replace a human driver. Tesla began rolling out an experimental version of that mode in October 2020. Although it requires drivers to keep their hands on the wheel, it currently offers such features as the ability to slow a car down automatically when it approaches stop signs or traffic lights. In February, Tesla recalled more than 362,000 US vehicles to update their full self-driving software after the National Highway Traffic Safety Administration said the software could allow vehicles to exceed speed limits and cause crashes at intersections. A driver died in a 2018 crash in California while riding in a Tesla vehicle operating on ‘autopilot’. Tesla said the driver was not using the feature properly. A driver died in a 2018 crash in California while riding in a Tesla vehicle operating on ‘autopilot’. Tesla said the driver was not using the feature properly. Photograph: AP As with many artificial-intelligence projects, to develop autopilot, Tesla hired data labelers to identify objects in images and videos to teach the system how to respond when the vehicle was on the road or parked. Tesla initially outsourced data labeling to a San Francisco-based nonprofit then known as Samasource, people familiar with the matter told Reuters. The organization had an office in Nairobi, Kenya, and specialized in offering training and employment opportunities to disadvantaged women and youth. In 2016, Samasource was providing about 400 workers there for Tesla, up from about an initial 20, according to a person familiar with the matter. By 2019, however, Tesla was no longer satisfied with the work of Samasource’s data labelers. At an event called Tesla AI Day in 2021, Andrej Karpathy, then senior director of AI at Tesla, said: “Unfortunately, we found very quickly that working with a third party to get datasets for something this critical was just not going to cut it. Honestly, the quality was not amazing.” A former Tesla employee said of the Samasource labelers: “They would highlight fire hydrants as pedestrians. They would miss objects all the time. Their skill level to draw boxes was very low.” Samasource, now called Sama, declined to comment on its work for Tesla. Tesla decided to bring data labeling in-house. “Over time, we’ve grown to more than a 1,000-person data labeling [organization] that is full of professional labelers who are working very closely with the engineers,” Karpathy said in his August 2021 presentation. Karpathy didn’t respond to requests for comment. Tesla’s own data labelers initially worked in the San Francisco Bay area, including the office in San Mateo. Groups of data labelers were assigned a variety of different tasks, including labeling street lane lines or emergency vehicles, ex-employees said. At one point, Teslas on autopilot were having difficulty backing out of garages and would get confused when encountering shadows or objects such as garden hoses. So some data labelers were asked to identify objects in videos recorded inside garages. The problem eventually was solved. In interviews, two former employees said in their normal work duties they were sometimes asked to view images of customers in and around their homes, including inside garages. “I sometimes wondered if these people know that we’re seeing that,” said one. “I saw some scandalous stuff sometimes, you know, like I did see scenes of intimacy but not nudity,” said another. “And there was just definitely a lot of stuff that like, I wouldn’t want anybody to see about my life.” As an example, this person recalled seeing “embarrassing objects”, such as “certain pieces of laundry, certain sexual wellness items and just private scenes of life that we really were privy to because the car was charging”. Memes in the San Mateo office Tesla staffed its San Mateo office with mostly young workers, in their 20s and early 30s, who brought with them a culture that prized entertaining memes and viral online content. Former staffers described a free-wheeling atmosphere in chatrooms with workers exchanging jokes about images they viewed while labeling. According to several ex-employees, some labelers shared screenshots, sometimes marked up using Adobe Photoshop, in private group chats on Mattermost, Tesla’s internal messaging system. There they would attract responses from other workers and managers. Participants would also add their own marked-up images, jokes or emojis to keep the conversation going. Some of the emojis were custom-created to reference office inside jokes, several ex-employees said. Knowing how much data those vehicles are capable of collecting definitely made folks nervous Former Tesla employee One former labeler described sharing images as a way to “break the monotony”. Another described how the sharing won admiration from peers. “If you saw something cool that would get a reaction, you post it, right, and then later, on break, people would come up to you and say, ‘Oh, I saw what you posted. That was funny,’” said the former labeler. “People who got promoted to lead positions shared a lot of these funny items and gained notoriety for being funny.” Some of the shared content resembled memes on the internet. There were dogs, interesting cars, and clips of people recorded by Tesla cameras tripping and falling. There was also disturbing content, such as someone being dragged into a car seemingly against their will, said one ex-employee. Video clips of crashes involving Teslas were also sometimes shared in private chats on Mattermost, several former employees said. Those included examples of people driving badly or collisions involving people struck while riding bikes, such as the one with the child, or motorcycles. Some data labelers would rewind such clips and play them in slow motion. At times, Tesla managers would crack down on inappropriate sharing of images on public Mattermost channels since they claimed the practice violated company policy. Still, screenshots and memes based on them continued to circulate through private chats on the platform, several ex-employees said. Workers shared them one-on-one or in small groups as recently as the middle of last year. One of the perks of working for Tesla as a data labeler in San Mateo was the chance to win a prize – the use of a company car for a day or two – according to two former employees. But some of the lucky winners became paranoid when driving the electric cars. “Knowing how much data those vehicles are capable of collecting definitely made folks nervous,” one ex-employee said.

1 0

seminario venerdì 21 aprile 2023 " La protezione dei dati personali e la transizione digitale: le sfide attuali e future"
by Mauro Alovisio April 7, 2023

April 7, 2023

Gent.mi ho il piacere di segnalare il seminario di presentazione della quinta edizione del corso di perfezionamento in materia di protezione dei dati personali dell'Università degli studi di Torino in sinergia con l'Unione Industriale di Torino*, *che si svolgerà *venerdì 21 aprile 2023 alle ore 17 presso l'aula magna del Campus Luigi Einaudi a Torino * *ad oggetto: "La protezione dei dati personali e la* *transizione digitale: le sfide attuali e future*" *.* Parteciperanno come relatori: il prof. *Francesco Pizzetti, il prof. Sergio Foà,* il *dott Agostino Ghiglia* dell'*Autorità Garante per la protezione* dei dati personali e rappresentanti dell'Unione Industriale e degli ordini professionali. Durante la presentazione saranno esaminate alcune delle problematiche di forte attualità di questi ultimi mesi: l'intelligenza artificiale, il ruolo del Garante, gli attacchi informatici e data breach, il trasferimento di dati extra Ue. Il seminario è accreditato per la formazione obbligatoria per gli avvocati Per informazioni (borse di studio under 29, calendario e scadenze) > > https://www.giurisprudenza.unito.it/do/home.pl/View?doc=data_protection_off… > Un caro saluto > Mauro Alovisio

1 0

30 under 30-year sentences: why so many of Forbes’ young heroes face jail
by Alberto Cammozzo April 7, 2023

April 7, 2023

<https://www.theguardian.com/business/2023/apr/06/forbes-30-under-30-tech-fi…> ust a few years ago, Charlie Javice was riding high. In 2019 the tech CEO landed a spot on Forbes’s 30 Under 30 list for her work on a startup called Frank, which she described as “Amazon for higher education”. What does that catchy but completely empty phrase mean? It means Frank helped students navigate the financial aid process. It was apparently so successfully at doing this that JPMorgan Chase acquired the company for $175m in 2021 and Javice was made a managing partner at the bank. [...] The problem here isn’t Forbes, of course; the problem is the vision of success that we’ve been sold and the fetishizing of youth. 30 Under 30 isn’t just a list, it’s a mentality: a pressure to achieve great things before youth slips away from you. The pressure can lead certain ambitious people to take shortcuts. And, in fact, shortcuts are encouraged: millennials, after all, grew up being told to “fake it ’til you make it”, cash in now until you become a withered, irrelevant, 30-year-old prune. If you exaggerate a little bit, that’s not fraud, that’s hustle! Until, of course, the justice department comes knocking.

1 0