project7

Faster sorting algorithms discovered using deep reinforcement learning

https://www.nature.com/articles/s41586-023-06004-9.pdf

Fundamental algorithms such as sorting or hashing are used trillions of times on any given day. As demand for computation grows, it has become critical for these algorithms to be as performant as possible. Whereas remarkable progress has been achieved in the past, making further improvements on the efficiency of these routines has proved challenging for both human scientists and computational approaches. Here we show how artificial intelligence can go beyond the current state of the art by discovering hitherto unknown routines. To realize this, we formulated the task of finding a better sorting routine as a single-player game. We then trained a new deep reinforcement learning agent, AlphaDev, to play this game. AlphaDev discovered small sorting algorithms from scratch that outperformed previously known human benchmarks. These algorithms have been integrated into the LLVM standard C++ sort library. This change to this part of the sort library represents the replacement of a component with an algorithm that has been automatically discovered using reinforcement learning. We also present results in extra domains, showcasing the generality of the approach.

complacency = compiacimento – sicurezza di sé oltre il ragionevole dubbio rispetto ai rischi che si possono incontrare

https://opensource.com/article/23/4/my-website-compromised by David Both [...] “What I learned Having written many books and articles in which I discuss the necessity to keep systems updated with the latest versions of operating system and application software, I'm frankly embarrassed by this. However, it has been a good learning experience for me and a reminder that I must not become complacent. I almost didn't write this article! I didn't want to admit to being negligent with one of my own systems. And yet, I felt compelled to write about it in the hope that you learn from my experience.

So, as I have learned from painful experience, it is critical to keep our systems updated. It's one of the most vital steps in the continuing battle to prevent the computers under our care from being infected. The specific details of the infection I experienced are less important than the fact that there are always attacks taking place against our systems. Complacency is one of the attack vectors that crackers can count on to aid their efforts.”

Essere troppo sicuri porta sempre a sottovalutare qualcosa in ambito personale, lavorativo, della vita in genere, al punto che qualcuno se ne può approfitare per trovare la falla, il momento di disattenzione. Nella maggior parte dei casi un buon “backup” non solo informatico è di estremo aiuto, alcune volte si prendo delle bastonate significative.

#tech

https://www.washingtonpost.com/technology/2023/06/02/ai-taking-jobs/

ChatGPT took their jobs. Now they walk dogs and fix air conditioners.

Ai took human job

...Roberts said that chatbots can produce costly errors and that companies rushing to incorporate ChatGPT into operations are “jumping the gun.” Since they work by predicting the most statistically likely word in a sentence, they churn out average content by design. That provides companies with a tough decision, she said: quality vs. cost.

“We have to ask: Is a facsimile good enough? Is imitation good enough? Is that all we care about?” she said. “We’re going to lower the measure of quality, and to what end? So the company owners and shareholders can take a bigger piece of the pie?”

Lipkin, the copywriter who discovered she’d been replaced by ChatGPT, is reconsidering office work altogether. She initially got into content marketing so that she could support herself while she pursued her own creative writing. But she found the job burned her out and made it hard to write for herself. Now, she’s starting a job as a dog walker.

Cit https://www.washingtonpost.com/technology/2023/04/05/chatgpt-lies/?itid=ap_pranshuverma&itid=lk_inline_manual_44

https://www.washingtonpost.com/wellness/2023/06/01/eating-disorder-chatbot-ai-weight-loss/?itid=lk_inline_manual_42

https://s3.documentcloud.org/documents/23826751/mata-v-avianca-airlines-affidavit-in-opposition-to-motion.pdf

https://www.washingtonpost.com/media/2023/01/17/cnet-ai-articles-journalism-corrections/?itid=lk_inline_manual_42

Avvenire https://www.avvenire.it/opinioni/pagine/senza-collaudi-le-chatbot-pi-dannose-dei-social

Elena Molinari scrive:

La massima istituzione medica statunitense (il Surgeon general USA) ha aggiunto l’uso costante di Facebook, Instagram e TikTok al fumo, alla guida in stato di ubriachezza e all’obesità nella lista dei «gravi pericoli per la salute e la sicurezza».

La fonte originale è questo doc https://www.hhs.gov/surgeongeneral/priorities/youth-mental-health/social-media/index.html

Si trovano due allegati: Una sintesi e l'avviso completo dal titolo:

Social Media and Youth Mental Health

This Advisory describes the current evidence on the impacts of social media on the mental health of children and adolescents. It states that we cannot conclude social media is sufficiently safe for children and adolescents and outlines immediate steps we can take to mitigate the risk of harm to children and adolescents.

Avvenire 30 maggio 2023 Mathias Risse https://www.avvenire.it/opinioni/pagine/le-superintelligenze-artificiali-possono-decidere-di-eliminarci

“Una volta che l’IA generale sarà più intelligente di noi, potrebbe produrre qualcosa di più intelligente di sé stessa, e così via, forse molto rapidamente. Quel momento è noto come la singolarità, un’esplosione di intelligenza che sarebbe probabilmente il più grande evento della storia umana.”

IA GENERALE approccio non specialistico per affrontare problemi imparando dagli errori

Quanto costano gli errori di un ia?! In termini: economici, sociali, ecologici, umani,

#tech

https://calamarim.medium.com/cassandra-crossing-cosa-fare-delle-scuse-di-jeff-528472e4f060

By cassandra – Marco Calamari

(167) — Le scuse di Amazon non bastano, non basta che l’azienda restituisca una manciata di dollari. Perché non promette di lasciare che gli utenti fruiscano a loro piacimento della cultura che comprano.

11 settembre 2009 — Una cosa inusuale è accaduta nei giorni scorsi: riguarda la storia infinita di Amazon e della scomparsa del libro “1984” dallo scaffale elettronico di alcuni acquirenti degli e-book di Amazon che l’avevano regolarmente acquistato e lo utilizzavano sul loro lettore di e-book Kindle.

Il CEO di Amazon Jeff Bezos, uno degli uomini più ricchi del mondo, ha scritto a tutti i clienti danneggiati una umilissima lettera aperta di scuse, contenente tra l’altro la promessa di restituire i dollaroni sonanti che loro avevano pagato sotto forma di assegno spedito a casa.

Cassandra si era già occupata della questione, ma avrebbe ancora qualche cosa da dire. Sì, perché da una lettura della lettera (assolutamente consigliata) potrebbe sembrare che la questione sia finita bene e che per una volta un megadirettore abbia riconosciuto davvero di aver sbagliato su tutti i fronti e abbia fatto la cosa giusta: per un attimo anche io ho preso questo abbaglio.

Non è così! Assolutamente non è così e probabilmente merita spiegare il perché, non essendo la cosa evidente.

Non si tratta solo del fatto che il perdono viene richiesto per cercare di recuperare una parte dell’immagine di azienda attenta e rispettosa dei suoi clienti.

Non è nemmeno perché vedi caso questa azione riparatrice è fatta alla vigilia dell’udienza di una class action promossa da un gruppo di tifosi di Orwell che se lo volevano leggere sul Kindle.

Nemmeno perché molta gente si è accorta che il libro “1984”, pagato a lei una decina di sonanti dollaroni, si può trovare su qualsiasi bancarella di libri usati (cartacei) per un paio di dollari, avendo anche il vantaggio che se il bancarellaio si pente non se lo può più riprendere.

No, Jeff, è ciò di cui lei non chiede perdono e che non promette di fare mai più che non le fa meritare il perdono dei suoi clienti.

Lei non promette di non farlo mai più, o che vigilerà perché nella sua azienda nessuno lo faccia.

Lei non si scusa di avere venduto un costoso prodotto (il Kindle) ai suoi clienti con una funzionalità censoria e nascosta incorporata, che la sua azienda potrà continuare ad utilizzare a sua discrezione.

Lei non riconosce che le limitazioni forzate che impone alla cultura che rivende sono un danno per i suoi clienti, e la partecipazione e l’avallo di un modello di business errato ed immorale che danneggia la cultura e la società.

Lei non promette di rimuovere questa funzionalità dal Kindle e nemmeno di non usarla mai più.

Lei non promette, pur essendo probabilmente il più grande fornitore di cultura al mondo, di rinnegare il modello di business dei DRM e della carestia culturale artificialmente indotta per alzare i prezzi ed i guadagni.

Le sue scuse sono insufficienti, sono funzionali a mantenere i suoi clienti in posizione di inferiorità e di subordinazione, sono solo una azione di convenienza.

Lei implicitamente conferma che continuerà ad abbracciare il Lato Oscuro di chi vuole spezzare il circolo virtuoso della cultura per il suo profitto.

Scuse respinte. Se le metta pure dove vuole, magari in una cornice sulla sua scrivania, e le mediti bene.

Originally published at punto-informatico.it

Licenza d’utilizzo: i contenuti di questo articolo, dove non diversamente indicato, sono sotto licenza Creative Commons Attribuzione — Condividi allo stesso modo 4.0 Internazionale (CC BY-SA 4.0), tutte le informazioni di utilizzo del materiale sono disponibili a questo link.

Asahi shibun english ver

https://www.asahi.com/sp/ajw/articles/14907186

A decade-long data breach in Toyota’s much-touted online service put some information on more than 2 million vehicles at risk, the Japanese automaker said Friday.

Spanning from January 2012 to April 2023, the problem with Toyota’s cloud-based Connected service pertains only to vehicles in Japan, said spokesperson Hideaki Homma.

The Connected service reminds owners to get maintenance checks and links to streaming entertainment and provides help during emergencies. It can call for help after a crash or locate a car that’s been stolen.

No issues arising from the breach have been reported so far.

Although there is no evidence any information was leaked, copied or misused due to the breach, the data at risk includes: the vehicle identification number, which is separate from the license plate; the location of the vehicle and at what time it was there; and video footage taken by the vehicle, known as the “drive recorder” in Japan.

Such information cannot be used to identify individual owners, according to Toyota Motor Corp., which makes the Prius hybrid and Lexus luxury models.

Vehicles belonging to about 2.15 million people have been affected, including those who used net services called G-Link, G-Book and Connected.

Toyota’s Connected service in Japan is operated by a subsidiary. Until recently, no one noticed outside access to such information should have been turned off, Homma said.

“We are so sorry to have caused such trouble to all the people,” he said.

The problem is a major embarrassment for Japan’s top automaker, which has built a reputation for quality and attention to detail.

Automakers worldwide are competing to differentiate model offerings with the latest technology to lure buyers.

The problem with the system has been fixed, Homma said, so it’s safe to continue driving Connect-enabled vehicles as usual, and there is no need to bring them in for repairs.

#databreach #dataprotection #tech

Promed mailing

Fabio Manganiello

https://mobile.slashdot.org/story/23/05/12/2051201/millions-of-mobile-phones-come-pre-infected-with-malware-say-researchers

That of mobile phones hardware and firmware is a very mature industry that still has many competitors.

What it means is that everybody must keep their prices as low as possible if they want somebody to buy the phones that mount their components. To the point that, for low-end Android phones, chip and firmware manufacturers have basically no profit margins.

So what do you do when you have nearly no profit margins and no way out of it? Simple: you create profit margins out of thin air by pre-installing spyware on the firmware itself. The phone comes out of the box with the spyware already installed.

There's apparently a big market down there with dozens of available plugins. From gathering data from your SMS texts, to sniffing Facebook sessions, to accessing your location data, to shoveling custom ads, up to selling “device time” to criminals for all kind of purposes (just like you can rent an AWS node on the fly for a few hours, criminals can also rent access to somebody's phones for slots of up to 5 minutes to do whatever they want with it).

If a phone (or a smartwatch, or a smart TV) looks too cheap to be true, it's because it is. Devices like the Fairphone or Librem may be a bit more expensive at parity of features, but you're paying the right not to have surveillance pre-installed on those devices.

#tech #soc

Fabio Manganiello

https://mobile.slashdot.org/story/23/05/12/2051201/millions-of-mobile-phones-come-pre-infected-with-malware-say-researchers

That of mobile phones hardware and firmware is a very mature industry that still has many competitors.

What it means is that everybody must keep their prices as low as possible if they want somebody to buy the phones that mount their components. To the point that, for low-end Android phones, chip and firmware manufacturers have basically no profit margins.

So what do you do when you have nearly no profit margins and no way out of it? Simple: you create profit margins out of thin air by pre-installing spyware on the firmware itself. The phone comes out of the box with the spyware already installed.

There's apparently a big market down there with dozens of available plugins. From gathering data from your SMS texts, to sniffing Facebook sessions, to accessing your location data, to shoveling custom ads, up to selling “device time” to criminals for all kind of purposes (just like you can rent an AWS node on the fly for a few hours, criminals can also rent access to somebody's phones for slots of up to 5 minutes to do whatever they want with it).

If a phone (or a smartwatch, or a smart TV) looks too cheap to be true, it's because it is. Devices like the Fairphone or Librem may be a bit more expensive at parity of features, but you're paying the right not to have surveillance pre-installed on those devices.

Quando una cosa è gratuita sei tu il prodotto.

#tech #soc