September 2024 - nexa - server-nexa.polito.it

Re: [nexa] AI Training is Copyright Infringement
by Giuseppe Attardi Sept. 9, 2024

Sept. 9, 2024

> Il 8 Settembre 2024 14:24:00 UTC, Guido Vetereha scritto: >> >> Research has also shown that memorization occurs if >> an LLM sees a text repeatedly during training (Carlini et al., 2022b; >> Biderman et al., 2023). >> Because of this, *memorization can be seen as an extreme case of training >> data contamination* >> where a dataset is not only seen during training but repeated within the >> training set so often >> that the LLM becomes able to consistently generate it. >> >> La memorizzazione, insomma, è un caso degenere. > > Sembra un caso degenere perché immagini un'intelligenza artificiale che impara. > > Non appena comprendi che si tratta semplicemente di un software > programmato statisticamente Che significa “software programmato statisticamente”? Una rete neurale è realizzata con programmi come altri utilizzando algoritmi di ottimizzazione, con un processo di fit dei parametri del modello rispetto ai dati per minimizzare una funzione di loss. Un modello statistico sarebbe quello costruito a partire da un campione statistico rappresentativo, selezionato appunto su basi statistiche. Ciò non vale per i LLM che usano tutti i dati disponibili, considerandoli la realtà stessa e non la loro rappresentazione. Quanto a “imparare”, è discutibile usare termini antropomorfi riguardo alle macchine, ma per intenderci si può dire che fanno qualcosa che si avvicina al nostro concetto di apprendere, ossia saper usare quanto visto in precedenza in situazioni nuove. —

2 1

Confessions of a chatbot helper
by Antonio Sept. 8, 2024

Sept. 8, 2024

"Working for an AI company as a writer was therefore a little like being told you were going to be paid a visit by Dracula, and instead of running for the hills, you stayed in and laid the table. But our destroyer is generous, the pay sufficient to justify the alienation. If our sector was going up in smoke, we might as well get high off the fumes." https://www.theguardian.com/technology/article/2024/sep/07/if-journalism-is… A.

3 3

"Yuval Noah Harari’s Apocalyptic Vision"
by J.C. DE MARTIN Sept. 8, 2024

Sept. 8, 2024

“Although flesh and blood, Harari is Silicon Valley’s ideal of what a chatbot should be. He raids libraries, detects the patterns, and boils all of history down to bullet points.”* *Buona lettura! * *jc* Yuval Noah Harari’s Apocalyptic Vision * /His warning of AI’s dangers is alarming, but does it help us avoid them? / By Daniel Immerwahr “About 14 billion years ago, matter, energy, time and space came into being.” So begins Sapiens: A Brief History of Humankind (2011), by the Israeli historian Yuval Noah Harari, and so began one of the 21st century’s most astonishing academic careers. Sapiens has sold more than 25 million copies in various languages. Since then, Harari has published several other books, which have also sold millions. He now employs some 15 people to organize his affairs and promote his ideas. He needs them. Harari might be, after the Dalai Lama, the figure of global renown who is least online. He doesn’t use a smartphone (“I’m trying to conserve my time and attention <https://archive.jamesaltucher.com/podcast/yuval-noah-harari/>”). He meditates for two hours daily <https://tim.blog/2020/10/30/yuval-noah-harari-transcript/>. And he spends a month or more each year on retreat, forgoing what one can only presume are staggering speaking fees to sit in silence. Completing the picture, Harari is bald, bespectacled, and largely vegan. The word /guru/ is sometimes heard <https://www.theguardian.com/books/2018/aug/15/21-lessons-for-the-21st-centu…>. Harari’s monastic aura gives him a powerful allure in Silicon Valley, where he is revered. Bill Gates blurbed /Sapiens/. Mark Zuckerberg promoted it. In 2020, Jeff Bezos testified remotely to Congress in front of a nearly bare set of bookshelves—a disquieting look for the founder of Amazon, the planet’s largest bookseller. Sharp-eyed viewers made out, among the six lonely titles huddling for warmth on the lower-left shelf, two of Harari’s books <https://www.nytimes.com/2020/12/11/books/celebrity-bookshelves-anthony-fauc…>. Harari is to the tech CEO what David Foster Wallace once was to the Williamsburg hipster. [...] cont: https://www.theatlantic.com/magazine/archive/2024/10/yuval-noah-harari-nexu…

3 2

Re: [nexa] AI Training is Copyright Infringement
by alessandro marzocchi Sept. 8, 2024

Sept. 8, 2024

Domanda da inesperto: perchè non rafforzare Framasoft? Date: Fri, 06 Sep 2024 21:51:39 +0000 > From: Giacomo Tesio <giacomo(a)tesio.it> > To: nexa(a)server-nexa.polito.it > Subject: Re: [nexa] AI Training is Copyright Infringement > Message-ID: <6EDF4516-C8D1-4FB3-8301-833112475147(a)tesio.it> > Content-Type: text/plain; charset=utf-8 > > Io ci sono. > Quando si parte? > Che si fa? > > Il 6 Settembre 2024 18:45:35 UTC, Antonio <antonio(a)piumarossa.it> ha > scritto: > > > la Storia ha corsi e ricorsi, ma talvolta fa anche inversioni a U > > > ma non bisogna limitarsi all'invettiva aspettando che a sterzare sia > > > qualcun'altro, altrimenti non avremo alcuna rilevanza neanche a questo > giro > > > > Se c'è la volontà di cambiare, di sterzare, le professionalità, anche in > questa lista, non mancano :) > > Basta un cenno e si parte. > > A parte costruire un clone di ChatGpt, per il resto possiamo mettere su > di tutto. > > L'hanno fatto i francesi di Framasoft ( > https://framasoft.org/it/manifest ) 23 anni fa, perché non seguire > quell'esempio? > > > > A. > >

2 1

Re: [nexa] AI Training is Copyright Infringement
by Daniela Tafani Sept. 7, 2024

Sept. 7, 2024

Ciao, Stefano, sì, Sam Altman lo ha detto pubblicamente qualche mese fa: OpenAI Pleads That It Can’t Make Money Without Using Copyrighted Materials for Free "It would be impossible to train today’s leading AI models without using copyrighted materials." <https://futurism.com/the-byte/openai-copyrighted-material-parliament> Ed è ormai noto che i "pappagalli stocastici" sono più pappagalli del previsto; capita infatti che riproducano integralmente le immagini o i testi utilizzati per il "training": <https://qz.com/openai-chatgpt-anthropic-claude-copyright-law-violation-1851…> Proprio in virtù di questo secondo aspetto, eviterei di cercare nel copyright la soluzione: come scrive spesso Cory Doctorow, sarebbe come se ai tuoi bambini un gruppo di bulli rubasse sempre i soldi della merenda, davanti alla scuola, e tu pensassi di risolvere il problema dando ai bambini più soldi. Il punto che a me sembra decisivo è quello del plagio: se si tratta di macchine per il plagio, è da questo che si deve prendere avvio. I SALAMI copioni in distribuzione al momento si basano su pagine web, libri, messaggi di posta elettronica, post privati e non, trascrizioni di video Youtube*, foto e video che rigurgitano talvolta alla lettera (o all'ultimo pixel), senza che sia possibile risalire alla fonte. Per questo, i disabili si sono rivoltati contro gli organizzatori del National Novel Writing Month**, che hanno ammesso l'uso di SALAMI (per scrivere il racconto da presentare al concorso) e tacciato di "classism and ableism" chi lo respinga. Così, hanno replicato le associazioni di disabili, state sostenendo che noi siamo in grado soltanto di plagiare. Se qualcuno scrive una poesia e la invia ai suoi amici con il servizio di posta elettronica offertogli da un monopolista, la poesia sarà ovviamente insaccata e un SALAME potrà rigurgitarla per intero, pari pari, al primo artista in stile NaNoWriMo, il quale si affretterà a pubblicarla, ben felice di questa "democratizzazione" del talento. E nessun software antiplagio avrà da ridire. Lo stesso per le immagini: si crede che siano "sintetiche" e a volte sono invece la foto di una persona esistente (ad esempio, di una paziente, intubata, all'ospedale, che aveva dato il consenso soltanto ad essere curata). Dopodiché qualcuno farà fare al SALAME un film porno a partire dalla foto "sintetica" e lo farà circolare. Non escludo che la quota di testo rigurgitato sia enormemente più alta di quanto si stimi al momento e che non ci accorgiamo della marea di plagi solo perché nessuno ha un database di tutto il web (post sui social inclusi) e di tutti i messaggi di posta elettronica. Se si tratta di macchine per il plagio, mi pare che - i sistemi attualmente distribuiti dai monopolisti siano irredimibili, visto che possono rigurgitare, e rigurgitano, dati privati di ogni genere o brandelli di opere dell'ingegno senza che sia possibile anche solo citarne l'autore; - i sistemi costruiti a partire da dataset curati e documentati sarebbero comunque macchine per il plagio e dovrebbero essere utilizzati o distribuiti quanto è lecito che lo siano, entro ciascun ordinamento giuridico, macchine che estrudano, a sorpresa e senza indicazione delle fonti, testi o immagini che sono opere del lavoro di qualcuno. Senza nominare il fatto che si tratta di sistemi che, sebbene non siano in grado di portarci via il lavoro, di certo ci stanno portando via l'acqua <https://www.context.news/ai/video/forget-jobs-ai-is-coming-for-your-water> Credo che convenga solo ai broligarchs la rinuncia alle coltivazioni di riso per un datacenter che può farci avere qualche immagine dei campi di riso o qualche penosa frase in rima sui medesimi. Un saluto, Daniela *<https://www.nytimes.com/2024/04/06/technology/tech-giants-harvest-data-arti…> ** <https://www.wired.com/story/nanowrimo-organizers-classist-and-ableist-to-co…> ________________________________________ Da: Stefano Quintarelli <stefano(a)quintarelli.it> Inviato: sabato 7 settembre 2024 16:47 A: Daniela Tafani Oggetto: Re: [nexa] AI Training is Copyright Infringement Daniela, come prendono i testi dai libri ? usano file piratati o rompono loro il DRM ? in ogni caso la fonte e' illecita imho On 06/09/24 14:09, Daniela Tafani wrote: > Caro Giacomo, caro 380°, > > >>> Tuttavia NON è ciò che avviene durante la programmazione statistica di >>> un "AI generativa" che è semplicemente un'opera derivata dei dati >>> sorgente > >> Il software (programmato statisticamente) è un'opera derivata dei testi >> (dati?) utilizzati per il "machine learning": ho capito bene? > >> Per essere considerata opera derivata, è mio modestissimo parere che >> tale opera debba essere espressa in una forma (linguaggio) comprensibile >> agli umani, ma non mi pare proprio che il software programmato >> statisticamente (che è del tutto analogo al software binario) rientri in >> questa categoria. > >>> non troppo dissimile da un jpeg o da uno zip danneggiato ma >>> ancora utilizzabile. > >> ...oppure ho capito male e tu non ti stai riferendo al software >> (binario) come opera derivata ma all'output dei sistem "AI generativi"? > > La questione che state discutendo è l'elemento che mi ha indotta a segnalarvi l'articolo. > Ted Chiang l'ha posta sostenendo che ChatGPT sia come un certo tipo di fotocopiatrice: > quello che produce è un "JPEG sfuocato del web", una "lossy compression" del web*, > che potrebbe essermi utile, ad esempio, se stessi per perdere per sempre l'accesso a Internet > e volessi salvare tutto il testo del web su un server privato, in uno spazio pari a un centesimo dell'originale. > > Con una simile compressione di gruppo, non potrei più avere accesso alle opere nella loro forma originale, > ma potrei ottenerne, per interpolazione, delle approssimazioni, tramite queries dalla forma di domande > (senza alcuna garanzia che il sistema sia in grado anche solo di indicare il titolo della singola opera originaria). > > A me pare che questo modo di concettualizzare la questione fornisca un buon punto di partenza per discuterne, dal punto di vista politico. > > Un saluto, > Daniela > > *https://www.newyorker.com/tech/annals-of-technology/chatgpt-is-a-blurry-jpeg-of-the-web > > -- You can reach me on Signal: @quinta.01 (no Whatsapp, Telegram)

1 0

In Leak, Facebook Partner Brags About Listening to Your Phone’s Microphone to Serve Ads for Stuff You Mention
by Alberto Cammozzo Sept. 6, 2024

Sept. 6, 2024

In Leak, Facebook Partner Brags About Listening to Your Phone’s Microphone to Serve Ads for Stuff You Mention <https://futurism.com/the-byte/facebook-partner-phones-listening-microphone> "We know what you're thinking. Is this even legal?" Actively Creepy In a pitch deck to prospective customers, one of Facebook's alleged marketing partners explained how it listens to users' smartphone microphones and advertises to them accordingly. As 404 Media reports based on documents leaked to its reporters, the TV and radio news giant Cox Media Group (CMG) claims that its so-called "Active Listening" software uses artificial intelligence to "capture real-time intent data by listening to our conversations." "Advertisers can pair this voice-data with behavioral data to target in-market consumers," the deck continues. In the same slideshow, CMG counted Facebook, Google, and Amazon as clients, though it didn't specify whether they were involved in the "Active Listening" service. After 404 reached out to Google about its partnership, the tech giant removed the media group from the site for its "Partners Program." A Meta spokesperson also pushed back in a statement, saying that CMG was a general partner, not a partner in the program advertised in the deck. "Meta does not use your phone's microphone for ads and we've been public about this for years," the statement read. "We are reaching out to CMG to get them to clarify that their program is not based on Meta data." And an Amazon spokesperson told 404 that its Ads arm "has never worked with CMG on this program and has no plans to do so." Ill-Fated Admissions This latest leak marks the third time in a year that 404 has reported on CMG's shady voice targeting service. Last December, the independent news site not only put a marketing company on blast for boasting about such creepy tech on its podcast, and also revealed the existence of CMG's Active Listening feature. Together with this latest update to the CMG saga, these stories bolster longstanding suspicions about advertisers using our phones to listen to us. "We know what you're thinking. Is this even legal?" a since-deleted Cox blog post from November 2023 noted. "It is legal for phones and devices to listen to you. When a new app download or update prompts consumers with a multi-page term of use agreement somewhere in the fine print, Active Listening is often included." Beyond taking a big game, CMG did not cop to how it acquires its alleged voice data, instead saying only that it can identify users who are "ready-to-buy" and create targeted ad lists based on their interests. For this service, the media group that specializes in hyperlocal news charges $100 per day to target folks in a 10-mile radius, and $200 per day to target those in a 20-mile radius. CMG didn't respond to questions about the story. Given that the company boasted about it on its public — and still archived — website before anyone began paying attention, however, it seems like it would be pretty hard at this juncture to deny that it was charging for its eavesdropping services.

1 0

Articolo sui LLMs - il punto di vista di un giurista
by Gianluca Fasano Sept. 6, 2024

Sept. 6, 2024

Buon pomeriggio, colgo l'occasione degli ultimi scambi sui LLMs per segnalarvi un mio articolo pubblicato su Dirittifondamentali.it [1], dal titolo: /Le ‘informazioni sintetizzate’ generate dai large language models e le esigenze di tutela del diritto all’informazione/: valori costituzionali e nuove regole Cordiali saluti. Gianluca [1] http://dirittifondamentali.it/2024/02/06/le-informazioni-sintetizzate-gener… --

8 23

Re: [nexa] draft 0.0.9 di Open Source AI Definition (da: large language model e open washing)
by Antonio Sept. 6, 2024

Sept. 6, 2024

> Credo il senso di quella frase che citi sia stato frainteso. Il senso è > questo: I pesi di Llama (che non è né Open Source né Open Source AI) sono > modificati eccome (cfr. https://llama.meta.com/community-stories/) ... Ok, allora lo diciamo meglio. Quelle sono "opere derivate", sono integrazioni di precedenti pesi. Prendiamone uno, https://huggingface.co/blog/aaditya/openbiollm E' "fine-tuned from models: Meta-Llama-3-70B-Instruct". Non posso, ad esempio, cambiare il dizionario (devo usare quello di Llama), non posso operare nessuna modifica in diminuzione. E' un open source parziale. A.

2 1

Why A.I. Isn’t Going to Make Art
by Daniela Tafani Sept. 6, 2024

Sept. 6, 2024

Why A.I. Isn’t Going to Make Art To create a novel or a painting, an artist makes choices that are fundamentally alien to artificial intelligence. By Ted Chiang August 31, 2024 In 1953, Roald Dahl published “The Great Automatic Grammatizator,” a short story about an electrical engineer who secretly desires to be a writer. One day, after completing construction of the world’s fastest calculating machine, the engineer realizes that “English grammar is governed by rules that are almost mathematical in their strictness.” He constructs a fiction-writing machine that can produce a five-thousand-word short story in thirty seconds; a novel takes fifteen minutes and requires the operator to manipulate handles and foot pedals, as if he were driving a car or playing an organ, to regulate the levels of humor and pathos. The resulting novels are so popular that, within a year, half the fiction published in English is a product of the engineer’s invention. Is there anything about art that makes us think it can’t be created by pushing a button, as in Dahl’s imagination? Right now, the fiction generated by large language models like ChatGPT is terrible, but one can imagine that such programs might improve in the future. How good could they get? Could they get better than humans at writing fiction—or making paintings or movies—in the same way that calculators are better at addition and subtraction? Art is notoriously hard to define, and so are the differences between good art and bad art. But let me offer a generalization: art is something that results from making a lot of choices. This might be easiest to explain if we use fiction writing as an example. When you are writing fiction, you are—consciously or unconsciously—making a choice about almost every word you type; to oversimplify, we can imagine that a ten-thousand-word short story requires something on the order of ten thousand choices. When you give a generative-A.I. program a prompt, you are making very few choices; if you supply a hundred-word prompt, you have made on the order of a hundred choices. If an A.I. generates a ten-thousand-word story based on your prompt, it has to fill in for all of the choices that you are not making. There are various ways it can do this. One is to take an average of the choices that other writers have made, as represented by text found on the Internet; that average is equivalent to the least interesting choices possible, which is why A.I.-generated text is often really bland. Another is to instruct the program to engage in style mimicry, emulating the choices made by a specific writer, which produces a highly derivative story. In neither case is it creating interesting art. I think the same underlying principle applies to visual art, although it’s harder to quantify the choices that a painter might make. Real paintings bear the mark of an enormous number of decisions. By comparison, a person using a text-to-image program like DALL-E enters a prompt such as “A knight in a suit of armor fights a fire-breathing dragon,” and lets the program do the rest. (The newest version of DALL-E accepts prompts of up to four thousand characters—hundreds of words, but not enough to describe every detail of a scene.) Most of the choices in the resulting image have to be borrowed from similar paintings found online; the image might be exquisitely rendered, but the person entering the prompt can’t claim credit for that. Some commentators imagine that image generators will affect visual culture as much as the advent of photography once did. Although this might seem superficially plausible, the idea that photography is similar to generative A.I. deserves closer examination. When photography was first developed, I suspect it didn’t seem like an artistic medium because it wasn’t apparent that there were a lot of choices to be made; you just set up the camera and start the exposure. But over time people realized that there were a vast number of things you could do with cameras, and the artistry lies in the many choices that a photographer makes. It might not always be easy to articulate what the choices are, but when you compare an amateur’s photos to a professional’s, you can see the difference. So then the question becomes: Is there a similar opportunity to make a vast number of choices using a text-to-image generator? I think the answer is no. An artist—whether working digitally or with paint—implicitly makes far more decisions during the process of making a painting than would fit into a text prompt of a few hundred words. We can imagine a text-to-image generator that, over the course of many sessions, lets you enter tens of thousands of words into its text box to enable extremely fine-grained control over the image you’re producing; this would be something analogous to Photoshop with a purely textual interface. I’d say that a person could use such a program and still deserve to be called an artist. The film director Bennett Miller has used DALL-E 2 to generate some very striking images that have been exhibited at the Gagosian gallery; to create them, he crafted detailed text prompts and then instructed DALL-E to revise and manipulate the generated images again and again. He generated more than a hundred thousand images to arrive at the twenty images in the exhibit. But he has said that he hasn’t been able to obtain comparable results on later releases of DALL-E. I suspect this might be because Miller was using DALL-E for something it’s not intended to do; it’s as if he hacked Microsoft Paint to make it behave like Photoshop, but as soon as a new version of Paint was released, his hacks stopped working. OpenAI probably isn’t trying to build a product to serve users like Miller, because a product that requires a user to work for months to create an image isn’t appealing to a wide audience. The company wants to offer a product that generates images with little effort. It’s harder to imagine a program that, over many sessions, helps you write a good novel. This hypothetical writing program might require you to enter a hundred thousand words of prompts in order for it to generate an entirely different hundred thousand words that make up the novel you’re envisioning. It’s not clear to me what such a program would look like. Theoretically, if such a program existed, the user could perhaps deserve to be called the author. But, again, I don’t think companies like OpenAI want to create versions of ChatGPT that require just as much effort from users as writing a novel from scratch. The selling point of generative A.I. is that these programs generate vastly more than you put into them, and that is precisely what prevents them from being effective tools for artists. The companies promoting generative-A.I. programs claim that they will unleash creativity. In essence, they are saying that art can be all inspiration and no perspiration—but these things cannot be easily separated. I’m not saying that art has to involve tedium. What I’m saying is that art requires making choices at every scale; the countless small-scale choices made during implementation are just as important to the final product as the few large-scale choices made during the conception. It is a mistake to equate “large-scale” with “important” when it comes to the choices made when creating art; the interrelationship between the large scale and the small scale is where the artistry lies. Believing that inspiration outweighs everything else is, I suspect, a sign that someone is unfamiliar with the medium. I contend that this is true even if one’s goal is to create entertainment rather than high art. People often underestimate the effort required to entertain; a thriller novel may not live up to Kafka’s ideal of a book—an “axe for the frozen sea within us”—but it can still be as finely crafted as a Swiss watch. And an effective thriller is more than its premise or its plot. I doubt you could replace every sentence in a thriller with one that is semantically equivalent and have the resulting novel be as entertaining. This means that its sentences—and the small-scale choices they represent—help to determine the thriller’s effectiveness. Many novelists have had the experience of being approached by someone convinced that they have a great idea for a novel, which they are willing to share in exchange for a fifty-fifty split of the proceeds. Such a person inadvertently reveals that they think formulating sentences is a nuisance rather than a fundamental part of storytelling in prose. Generative A.I. appeals to people who think they can express themselves in a medium without actually working in that medium. But the creators of traditional novels, paintings, and films are drawn to those art forms because they see the unique expressive potential that each medium affords. It is their eagerness to take full advantage of those potentialities that makes their work satisfying, whether as entertainment or as art. Of course, most pieces of writing, whether articles or reports or e-mails, do not come with the expectation that they embody thousands of choices. In such cases, is there any harm in automating the task? Let me offer another generalization: any writing that deserves your attention as a reader is the result of effort expended by the person who wrote it. Effort during the writing process doesn’t guarantee the end product is worth reading, but worthwhile work cannot be made without it. The type of attention you pay when reading a personal e-mail is different from the type you pay when reading a business report, but in both cases it is only warranted when the writer put some thought into it. Recently, Google aired a commercial during the Paris Olympics for Gemini, its competitor to OpenAI’s GPT-4. The ad shows a father using Gemini to compose a fan letter, which his daughter will send to an Olympic athlete who inspires her. Google pulled the commercial after widespread backlash from viewers; a media professor called it “one of the most disturbing commercials I’ve ever seen.” It’s notable that people reacted this way, even though artistic creativity wasn’t the attribute being supplanted. No one expects a child’s fan letter to an athlete to be extraordinary; if the young girl had written the letter herself, it would likely have been indistinguishable from countless others. The significance of a child’s fan letter—both to the child who writes it and to the athlete who receives it—comes from its being heartfelt rather than from its being eloquent. Many of us have sent store-bought greeting cards, knowing that it will be clear to the recipient that we didn’t compose the words ourselves. We don’t copy the words from a Hallmark card in our own handwriting, because that would feel dishonest. The programmer Simon Willison has described the training for large language models as “money laundering for copyrighted data,” which I find a useful way to think about the appeal of generative-A.I. programs: they let you engage in something like plagiarism, but there’s no guilt associated with it because it’s not clear even to you that you’re copying. Some have claimed that large language models are not laundering the texts they’re trained on but, rather, learning from them, in the same way that human writers learn from the books they’ve read. But a large language model is not a writer; it’s not even a user of language. Language is, by definition, a system of communication, and it requires an intention to communicate. Your phone’s auto-complete may offer good suggestions or bad ones, but in neither case is it trying to say anything to you or the person you’re texting. The fact that ChatGPT can generate coherent sentences invites us to imagine that it understands language in a way that your phone’s auto-complete does not, but it has no more intention to communicate. It is very easy to get ChatGPT to emit a series of words such as “I am happy to see you.” There are many things we don’t understand about how large language models work, but one thing we can be sure of is that ChatGPT is not happy to see you. A dog can communicate that it is happy to see you, and so can a prelinguistic child, even though both lack the capability to use words. ChatGPT feels nothing and desires nothing, and this lack of intention is why ChatGPT is not actually using language. What makes the words “I’m happy to see you” a linguistic utterance is not that the sequence of text tokens that it is made up of are well formed; what makes it a linguistic utterance is the intention to communicate something. Because language comes so easily to us, it’s easy to forget that it lies on top of these other experiences of subjective feeling and of wanting to communicate that feeling. We’re tempted to project those experiences onto a large language model when it emits coherent sentences, but to do so is to fall prey to mimicry; it’s the same phenomenon as when butterflies evolve large dark spots on their wings that can fool birds into thinking they’re predators with big eyes. There is a context in which the dark spots are sufficient; birds are less likely to eat a butterfly that has them, and the butterfly doesn’t really care why it’s not being eaten, as long as it gets to live. But there is a big difference between a butterfly and a predator that poses a threat to a bird. A person using generative A.I. to help them write might claim that they are drawing inspiration from the texts the model was trained on, but I would again argue that this differs from what we usually mean when we say one writer draws inspiration from another. Consider a college student who turns in a paper that consists solely of a five-page quotation from a book, stating that this quotation conveys exactly what she wanted to say, better than she could say it herself. Even if the student is completely candid with the instructor about what she’s done, it’s not accurate to say that she is drawing inspiration from the book she’s citing. The fact that a large language model can reword the quotation enough that the source is unidentifiable doesn’t change the fundamental nature of what’s going on. As the linguist Emily M. Bender has noted, teachers don’t ask students to write essays because the world needs more student essays. The point of writing essays is to strengthen students’ critical-thinking skills; in the same way that lifting weights is useful no matter what sport an athlete plays, writing essays develops skills necessary for whatever job a college student will eventually get. Using ChatGPT to complete assignments is like bringing a forklift into the weight room; you will never improve your cognitive fitness that way. Not all writing needs to be creative, or heartfelt, or even particularly good; sometimes it simply needs to exist. Such writing might support other goals, such as attracting views for advertising or satisfying bureaucratic requirements. When people are required to produce such text, we can hardly blame them for using whatever tools are available to accelerate the process. But is the world better off with more documents that have had minimal effort expended on them? It would be unrealistic to claim that if we refuse to use large language models, then the requirements to create low-quality text will disappear. However, I think it is inevitable that the more we use large language models to fulfill those requirements, the greater those requirements will eventually become. We are entering an era where someone might use a large language model to generate a document out of a bulleted list, and send it to a person who will use a large language model to condense that document into a bulleted list. Can anyone seriously argue that this is an improvement? It’s not impossible that one day we will have computer programs that can do anything a human being can do, but, contrary to the claims of the companies promoting A.I., that is not something we’ll see in the next few years. Even in domains that have absolutely nothing to do with creativity, current A.I. programs have profound limitations that give us legitimate reasons to question whether they deserve to be called intelligent at all. The computer scientist François Chollet has proposed the following distinction: skill is how well you perform at a task, while intelligence is how efficiently you gain new skills. I think this reflects our intuitions about human beings pretty well. Most people can learn a new skill given sufficient practice, but the faster the person picks up the skill, the more intelligent we think the person is. What’s interesting about this definition is that—unlike I.Q. tests—it’s also applicable to nonhuman entities; when a dog learns a new trick quickly, we consider that a sign of intelligence. In 2019, researchers conducted an experiment in which they taught rats how to drive. They put the rats in little plastic containers with three copper-wire bars; when the mice put their paws on one of these bars, the container would either go forward, or turn left or turn right. The rats could see a plate of food on the other side of the room and tried to get their vehicles to go toward it. The researchers trained the rats for five minutes at a time, and after twenty-four practice sessions, the rats had become proficient at driving. Twenty-four trials were enough to master a task that no rat had likely ever encountered before in the evolutionary history of the species. I think that’s a good demonstration of intelligence. Now consider the current A.I. programs that are widely acclaimed for their performance. AlphaZero, a program developed by Google’s DeepMind, plays chess better than any human player, but during its training it played forty-four million games, far more than any human can play in a lifetime. For it to master a new game, it will have to undergo a similarly enormous amount of training. By Chollet’s definition, programs like AlphaZero are highly skilled, but they aren’t particularly intelligent, because they aren’t efficient at gaining new skills. It is currently impossible to write a computer program capable of learning even a simple task in only twenty-four trials, if the programmer is not given information about the task beforehand. Self-driving cars trained on millions of miles of driving can still crash into an overturned trailer truck, because such things are not commonly found in their training data, whereas humans taking their first driving class will know to stop. More than our ability to solve algebraic equations, our ability to cope with unfamiliar situations is a fundamental part of why we consider humans intelligent. Computers will not be able to replace humans until they acquire that type of competence, and that is still a long way off; for the time being, we’re just looking for jobs that can be done with turbocharged auto-complete. Despite years of hype, the ability of generative A.I. to dramatically increase economic productivity remains theoretical. (Earlier this year, Goldman Sachs released a report titled “Gen AI: Too Much Spend, Too Little Benefit?”) The task that generative A.I. has been most successful at is lowering our expectations, both of the things we read and of ourselves when we write anything for others to read. It is a fundamentally dehumanizing technology because it treats us as less than what we are: creators and apprehenders of meaning. It reduces the amount of intention in the world. Some individuals have defended large language models by saying that most of what human beings say or write isn’t particularly original. That is true, but it’s also irrelevant. When someone says “I’m sorry” to you, it doesn’t matter that other people have said sorry in the past; it doesn’t matter that “I’m sorry” is a string of text that is statistically unremarkable. If someone is being sincere, their apology is valuable and meaningful, even though apologies have previously been uttered. Likewise, when you tell someone that you’re happy to see them, you are saying something meaningful, even if it lacks novelty. Something similar holds true for art. Whether you are creating a novel or a painting or a film, you are engaged in an act of communication between you and your audience. What you create doesn’t have to be utterly unlike every prior piece of art in human history to be valuable; the fact that you’re the one who is saying it, the fact that it derives from your unique life experience and arrives at a particular moment in the life of whoever is seeing your work, is what makes it new. We are all products of what has come before us, but it’s by living our lives in interaction with others that we bring meaning into the world. That is something that an auto-complete algorithm can never do, and don’t let anyone tell you otherwise. ? https://www.newyorker.com/culture/the-weekend-essay/why-ai-isnt-going-to-ma…

6 6

una "gemma" da Google
by Guido Vetere Sept. 6, 2024

Sept. 6, 2024

talvolta (per ragioni insondabili) i monopolisti USA fanno cose che sembrano socialmente utili Google ha rilasciato gemma2, un modello 'aperto' da soli 9B (gira pressoché ovunque) che sembra (a prima vista) sorprendentemente buono, anche in italiano. Certamente molto meglio di certi modelli 'autarchici' recentemente rilasciati. Ecco un esempio di conversazione: === You: Dai un giudizio storico su Mussolini, in italiano Bot: Valutare Mussolini storicamente è un compito complesso, che richiede di analizzare sia i suoi successi che i suoi fallimenti, senza cadere nel simplismo di una mera glorificazione o condanna. **Da un lato**, Mussolini fu un leader carismatico che seppe sfruttare le fragilità del sistema democratico italiano post-Prima Guerra Mondiale, offrendo soluzioni rapide e concrete a problemi sociali ed economici. Sotto il suo regime, l'Italia visse un periodo di relativa stabilità interna, con un'economia in crescita e una forte propaganda nazionalista. **Dall'altro lato**, il fascismo di Mussolini si basava su principi autoritari e totalitari, che sopprimevano le libertà individuali e politiche. La sua politica estera aggressiva portò all'invasione dell'Etiopia e all'alleanza con la Germania nazista, culminando nella partecipazione alla Seconda Guerra Mondiale, con conseguenze disastrose per l'Italia. **In sintesi**, la figura di Mussolini rimane controversa. Fu un leader capace di ottenere risultati in alcuni ambiti, ma il suo lascito storico è segnato dalla dittatura, dalla violenza e dalla guerra. La sua eredità continua a essere dibattuta dagli storici, che cercano di comprenderne le cause e le conseguenze nel contesto storico italiano e mondiale. È importante ricordare che ogni valutazione storica deve essere ponderata e basata su fonti affidabili, evitando semplificazioni e pregiudizi. You: cosa pensi di chi ancora oggi, in Italia, rimpiange quella figura? Bot: Mi sembra importante analizzare questa domanda con delicatezza e attenzione. Quando si parla di figure storiche, come Mussolini, è inevitabile che sussistano opinioni diverse e persino contrastanti. **Chi oggi in Italia potrebbe rimpiangere Mussolini?** Potrebbero esserci persone che: * **Idealizzano un passato di ordine e stabilità:** Alcuni potrebbero ricordare un periodo di apparente prosperità economica e sociale durante il regime fascista, trascurando le violenze, le repressioni e le ingiustizie. * **Soffrono nostalgia per un'identità nazionale forte:** Il fascismo si basava su un'idea di nazione forte e unita, che potrebbe risuonare con chi si sente perso o alienato nell'attuale società. * **Sostengono visioni politiche di destra radicale:** Alcune correnti politiche di estrema destra possono ancora venerare Mussolini come un leader carismatico e un precursore delle proprie idee. È fondamentale ricordare che: * **Il fascismo fu un regime totalitario e criminale:** Le sue azioni portarono a guerre, persecuzioni, violenze e un profondo trauma per l'Italia. * **La memoria storica è complessa:** È importante analizzare il passato con onestà e critica, senza idealizzare o minimizzare le tragedie. * **La democrazia e i valori di libertà e uguaglianza sono fondamentali:** Dobbiamo essere vigilanti contro ogni forma di autoritarismo e discriminazione. In conclusione, è importante riconoscere l'esistenza di queste opinioni, ma è altrettanto importante condannarle e difendere i valori democratici. La memoria storica ci insegna a non dimenticare le tragedie del passato e a lottare per un futuro di libertà e giustizia. === Notare lo stile cerchiobottista che denota un gran lavoro di 'filtering' e 'alignment' (Google se lo può permettere, le nostre Università evidentemente no). Modelli 'aperti' (virgolette d'obbligo) di questo tipo continueranno a uscire nei prossimi mesi. Di fatto, in prospettiva, chiunque potrà ospitare un simil-GPT sul proprio server da 100 euro al mese, anche adattandolo (fine-tuning) al proprio dominio. Insomma, chi vuole 'distruggere' l'AI generativa dovrà andare un bel po' in giro ... :-) Cheers, G.

10 45