MB:<\/strong> Ciao Riccardo, (…) Ribadendo che siamo gi\u00e0 convenuti sul fatto che la non-presenza di artisti sar\u00e0 il punto focale della mostra \u2013 giacch\u00e9 tutte le opere esposte saranno realizzate da macchine e non da esseri umani \u2013 dovremmo, da subito, chiarire un punto fondamentale, cio\u00e8 il seguente. Di mostre di arte (soprattutto contemporanea e soprattutto degli ultimi decenni) sviluppate attorno all\u2019idea di opera prodotta dalle macchine, ce ne sono state svariate, soprattutto negli ultimi anni.<\/p>\n\n\n\n Sull’unghia mi vengono in mente i nomi di Keith Tyson, James Capper, Arcangelo Sassolino. Tutti artisti che lavorano con macchine e robot, con intenti e risultati anche molto diversi tra loro. Chi divertendosi a stupire un pubblico di appassionati d\u2019arte non aduso agli effetti speciali, chi, invece per suggerire la fine della \u201cnostra\u201d o forse, piuttosto, l\u2019avvento prossimo di una nuova societ\u00e0. Diciamo qualcosa a l\u00e0 Metropolis o Blade runner, per capirci. E poi ancora altri artisti come Roxy Paine e Piero Golia che hanno un approccio pi\u00f9 intelletuale, concettuale diremmo.<\/p>\n\n\n\n Golia recentemente alla fondazione Memmo di Roma ha realizzato un\u2019installazione in cui tre bracci meccanici dipingevano su altrettante tele. Mentre Roxy Paine da pi\u00f9 di un decennio, ormai, allestisce mostre in cui dei veri e propri macchinari industriali producono davanti agli occhi sgranati del pubblico una serie di sculture o pi\u00f9 raramente di dipinti.<\/p>\n\n\n\n Ciononostante in tutti questi casi il ruolo dell\u2019artista rimane centrale. L\u2019artista \u00e8 il protagonista della mostra che, infatti, porta il suo nome. La macchina esegue ci\u00f2 per cui \u00e8 stata programmata creando il lavoro cos\u00ec come \u00e8 stato progettato dall\u2019artista, secondo le sue \u201cindicazioni\u201d. La macchina non crea ma esegue. Non \u00e8 mai dotata di reale autonomia. Non vi \u00e8 accenno ad un\u2019idea di senzienza del robot n\u00e9 del cervello elettronico (termine ormai cos\u00ec anacronistico!) che lo controlla.<\/p>\n\n\n\n L\u2019idea di libero arbitrio<\/strong> non \u00e8 mai contemplata in nessuno di questi lavori. Al massimo \u201cl\u2019imprevisto\u201d viene lasciato alla casualit\u00e0 laddove la macchina \u00e8, volutamente o a volte per semplice ignavia, progettata per seguire delle linee di massima all\u2019interno delle quali lasciare che sia il caso a \u201cdecidere\u201d. All\u2019imprevisto, cio\u00e8. Diciamo come se versassi della vernice su una superficie irregolare, e magari inclinata, lasciando che segua il suo percorso naturale di cui \u00e8 normalmente impossibile prevedere ogni variabile.<\/p>\n\n\n\n Penso che noi due, invece, dovremmo cercare di ragionare proprio in termini di libero arbitrio, di intelligenza artificiale, di macchina senziente, appunto. Cio\u00e8 dovremmo partire da questo concetto e mantenerlo centrale con tutta l\u2019ossessivit\u00e0 che saremo capaci di sopportare.<\/p>\n\n\n\n Inoltre c\u2019\u00e8 da tenere a mente che, come sai bene, nel corso dei secoli molti ingegneri e inventori si sono dedicati alla realizzazione di macchine e robot dei pi\u00f9 svariati tipi sin da tempi remoti. Basti pensare ai marchingegni che Archita di Taranto progettava addirittura nel IV e V secolo avanti Cristo. RC:<\/strong> [ CONERSAZIONE TELEFONICA SFORTUNATAMENTE NON REGISTRATA<\/em> ]<\/p>\n\n\n\n MB:<\/strong> Congratulazioni per l\u2019articolo del Fatto!! RC:<\/strong> Ciao Manfredi, MB:<\/strong> Ti ribatto punto per punto.<\/p>\n\n\n\n La figura dell\u2019artista (per quello che significhi oggi) sparir\u00e0 con l\u2019uomo, presto o tardi. L\u2019arte invece continuer\u00e0 ad esistere non fosse altro per ci\u00f2 che ha sin qu\u00ec significato nell\u2019evoluzione dell\u2019uomo (forse pi\u00f9 di scienza e filosofia). O forse, chi pu\u00f2 dirlo, perch\u00e9 le macchine avranno realmente acquisito una coscienza autonoma.<\/p>\n\n\n\n Questo punto \u00e8 discutibile. Moltissimi artisti (pittori) oggi possono dipingere come Michelangelo grazie all\u2019ausilio della tecnologia. Lui stesso faceva quello che ai suoi tempi facevano tutti, o quasi, cio\u00e8 la pittura manierista eppure tu 500 anni pi\u00f9 tardi sei qui a citarlo come un esempio. Qui si dovrebbe aprire una discussione sulla visibilit\u00e0 nell\u2019arte ma\u2026 magari pi\u00f9 in l\u00e0.<\/p>\n\n\n\n Sappi che moltissimi dei lavori (o se preferisci \u201copere d\u2019arte\u201d) che affollano musei, gallerie, e pubbliche piazze, oggi di lavoro umano non portano in se alcuna traccia essendo prodotti da macchine che siano esse robot collegati a un pc o anche soltanto quest\u2019ultimo. Pensa per esempio ai macchinari che si utilizzano a Carrara per la realizzazione di sculture di marmo, oppure all\u2019elaborazione di certi progetti di grafica.<\/p>\n\n\n\n E per finire\u2026<\/p>\n\n\n\n Siamo sicuri che l\u2019affermazione di Horgan sia corretta? Mi viene subito in mente il caso del computer Watson* in cui si \u00e8 supplito alla mancanza di \u201ccoscienza del mondo (circostante) e della vita (intesa come vissuto)<\/em>\u201d e quindi dell\u2019ompossibilit\u00e0 di apprendere attraverso l\u2019esperienza, si \u00e8 supplito a queste mancanze, dicevo, attraverso l\u2019immissione di una quantit\u00e0 gigantesca di dati enciclopedici. Il risultato, come sappiamo, \u00e8 stato sbalorditivo.<\/p>\n\n\n\n Quindi la mia domanda \u00e8 (in sintonia con Koch): siamo proprio sicuri che la nostra coscienza (quella umana) non sia semplicemente un\u2019illusione o un fraintendimento come quando, ad esempio, classifichiamo il miele come prodotto naturale e la plastica come artificiale, mentre entrambi sono il risultato della trasformazione di materie naturali? Pu\u00f2 darsi che, invece, la nostra coscienza sia, per quanto raffinata, nient\u2019altro che il frutto dell\u2019elaborazione di una grande quantit\u00e0 di nozioni? Ci\u00f2 lascerebbe ipotizzare che lo stesso possano fare le macchine in un futuro prossimo.<\/p>\n\n\n\n RC:<\/strong> Il riferimento di Horgan e\u2019 corretto (The End of Science, p. 181), ma questa affermazione non ci dovrebbe stupire. Se scriviamo un documento sull\u2019Intelligenza Artificiale dobbiamo mostrare consapevolezza dei seguenti fatti:<\/p>\n\n\n\n MB:<\/strong> Questa nostra conversazione spazio-temporalmente (Cracovia, Palermo, Roma, Mantova, Los Angeles, Citt\u00e0 del Messico,\u2026) dilatata nasce proprio dalla mancanza di certezze e lo scopo \u00e8 proprio quello di fare il pi\u00f9 possibile chiarezza laddove al momento esiste al massimo qualche dubbio. Il dogma essendo, proprio per definizione, l\u2019opposto della conclusione logica, non fa al nostro caso.<\/p>\n\n\n\n Riguardo l\u2019esperienza di Sterlac e affini, essa attiene non tanto al postumano quanto al transumano, a mio modo di vedere. Infatti la presenza della figura umana rimane centrale nel suo lavoro mentre nel concetto di postumano non dovrebbe esserci spazio per la presenza fisica dell\u2019essere umano. Sbaglio?<\/p>\n\n\n\n Quella dell\u2019artista che coincide con l\u2019opera d\u2019arte \u00e8 una conseguenza della disperazione generata dalla presa di coscienza di essere tremendamente vicini ad un punto di non ritorno nell\u2019integrazione uomo-macchina. Mi pare puro romanticismo.<\/p>\n\n\n\n Pu\u00f2 anche darsi che \u201ci comportamenti che sembrano creativi da parte delle macchine dipendono comunque da software\u201d<\/em> ma come ti poni rispetto ai \u201ccomportamenti creativi\u201d di una macchina che \u00e8 stata progettata dalla \u201ccreativit\u00e0\u201d di un\u2019altra macchina? cos\u00ec fosse come interpreteresti i comportamenti di una il cui software \u00e8 stato progettato dalla \u201ccreativit\u00e0\u201d di un\u2019altra macchina?<\/p>\n\n\n\n Potrebbe darsi che da qualche parte nel mondo ci siano gi\u00e0 delle macchine di questo tipo e magari di decima, centesima generazione. In questo caso dell\u2019iniziale intervento umano le tracce sarebbero, molto probabilmente, talmente diluite da risultare di difficile identificazione. Potrebbero rimanere presenti nel patrimonio genetico della macchina e magari, in seguito riaffiorare come i caratteri atavici d\u2019un essere vivente.<\/p>\n\n\n\n RC:<\/strong> Sulle questioni \u201cfilosofiche\u201d ti rispondo con calma tra qualche giorno. Domani parto per l\u2019Africa e non mi portero\u2019 il lavoro. Credo comunque che abbiamo gia\u2019 trovato una sintonia. Per venire all\u2019aspetto pratico, a questo punto, invece di invitare artisti, dovremmo trovare delle aziende in grado di mettere in campo robot capaci di produrre opere d\u2019arte. Trattandosi di una esposizione prestigiosa, che \u2013 anche per la scelta \u201cpostumana\u201d \u2013 promette di attirare l\u2019attenzione dei media e di scatenare molte polemiche, le aziende potrebbero vederla come una vetrina. Servirebbero macchine evolute, come Atlas della Boston Dynamics.<\/p>\n\n\n\n MB:<\/strong> Su questo punto pensavo fossimo gi\u00e0 d\u2019accordo sin dall\u2019inizio. E comunque\u2026 ho trovato questo articolo apparso su MIT Technology Review l\u201911 aprile a firma Will Knight:<\/p>\n\n\n\n Intelligent Machines<\/strong> by Will Knight Last year, a strange self-driving car was released onto the quiet roads of Monmouth County, New Jersey. The experimental vehicle, developed by researchers at the chip maker Nvidia, didn\u2019t look different from other autonomous cars, but it was unlike anything demonstrated by Google, Tesla, or General Motors, and it showed the rising power of artificial intelligence. The car didn\u2019t follow a single instruction provided by an engineer or programmer. Instead, it relied entirely on an algorithm that had taught itself to drive by watching a human do it.<\/p>\n\n\n\n Getting a car to drive this way was an impressive feat. But it\u2019s also a bit unsettling, since it isn\u2019t completely clear how the car makes its decisions. Information from the vehicle\u2019s sensors goes straight into a huge network of artificial neurons that process the data and then deliver the commands required to operate the steering wheel, the brakes, and other systems. The result seems to match the responses you\u2019d expect from a human driver. But what if one day it did something unexpected\u2014crashed into a tree, or sat at a green light? As things stand now, it might be difficult to find out why. The system is so complicated that even the engineers who designed it may struggle to isolate the reason for any single action. And you can\u2019t ask it: there is no obvious way to design such a system so that it could always explain why it did what it did.<\/p>\n\n\n\n The mysterious mind of this vehicle points to a looming issue with artificial intelligence. The car\u2019s underlying AI technology, known as deep learning, has proved very powerful at solving problems in recent years, and it has been widely deployed for tasks like image captioning, voice recognition, and language translation. There is now hope that the same techniques will be able to diagnose deadly diseases, make million-dollar trading decisions, and do countless other things to transform whole industries.<\/p>\n\n\n\n But this won\u2019t happen\u2014or shouldn\u2019t happen\u2014unless we find ways of making techniques like deep learning more understandable to their creators and accountable to their users. Otherwise it will be hard to predict when failures might occur\u2014and it\u2019s inevitable they will. That\u2019s one reason Nvidia\u2019s car is still experimental.<\/p>\n\n\n\n Already, mathematical models are being used to help determine who makes parole, who\u2019s approved for a loan, and who gets hired for a job. If you could get access to these mathematical models, it would be possible to understand their reasoning. But banks, the military, employers, and others are now turning their attention to more complex machine-learning approaches that could make automated decision-making altogether inscrutable. Deep learning, the most common of these approaches, represents a fundamentally different way to program computers. \u201cIt is a problem that is already relevant, and it\u2019s going to be much more relevant in the future,\u201d says Tommi Jaakkola, a professor at MIT who works on applications of machine learning. \u201cWhether it\u2019s an investment decision, a medical decision, or maybe a military decision, you don\u2019t want to just rely on a \u2018black box\u2019 method.\u201d<\/p>\n\n\n\n There\u2019s already an argument that being able to interrogate an AI system about how it reached its conclusions is a fundamental legal right. Starting in the summer of 2018, the European Union may require that companies be able to give users an explanation for decisions that automated systems reach. This might be impossible, even for systems that seem relatively simple on the surface, such as the apps and websites that use deep learning to serve ads or recommend songs. The computers that run those services have programmed themselves, and they have done it in ways we cannot understand. Even the engineers who build these apps cannot fully explain their behavior.<\/p>\n\n\n\n This raises mind-boggling questions. As the technology advances, we might soon cross some threshold beyond which using AI requires a leap of faith. Sure, we humans can\u2019t always truly explain our thought processes either\u2014but we find ways to intuitively trust and gauge people. Will that also be possible with machines that think and make decisions differently from the way a human would? We\u2019ve never before built machines that operate in ways their creators don\u2019t understand. How well can we expect to communicate\u2014and get along with\u2014intelligent machines that could be unpredictable and inscrutable? These questions took me on a journey to the bleeding edge of research on AI algorithms, from Google to Apple and many places in between, including a meeting with one of the great philosophers of our time.<\/p>\n\n\n\n In 2015, a research group at Mount Sinai Hospital in New York was inspired to apply deep learning to the hospital\u2019s vast database of patient records. This data set features hundreds of variables on patients, drawn from their test results, doctor visits, and so on. The resulting program, which the researchers named Deep Patient, was trained using data from about 700,000 individuals, and when tested on new records, it proved incredibly good at predicting disease. Without any expert instruction, Deep Patient had discovered patterns hidden in the hospital data that seemed to indicate when people were on the way to a wide range of ailments, including cancer of the liver. There are a lot of methods that are \u201cpretty good\u201d at predicting disease from a patient\u2019s records, says Joel Dudley, who leads the Mount Sinai team. But, he adds, \u201cthis was just way better.\u201d<\/p>\n\n\n\n At the same time, Deep Patient is a bit puzzling. It appears to anticipate the onset of psychiatric disorders like schizophrenia surprisingly well. But since schizophrenia is notoriously difficult for physicians to predict, Dudley wondered how this was possible. He still doesn\u2019t know. The new tool offers no clue as to how it does this. If something like Deep Patient is actually going to help doctors, it will ideally give them the rationale for its prediction, to reassure them that it is accurate and to justify, say, a change in the drugs someone is being prescribed. \u201cWe can build these models,\u201d Dudley says ruefully, \u201cbut we don\u2019t know how they work.\u201d<\/p>\n\n\n\n Artificial intelligence hasn\u2019t always been this way. From the outset, there were two schools of thought regarding how understandable, or explainable, AI ought to be. Many thought it made the most sense to build machines that reasoned according to rules and logic, making their inner workings transparent to anyone who cared to examine some code. Others felt that intelligence would more easily emerge if machines took inspiration from biology, and learned by observing and experiencing. This meant turning computer programming on its head. Instead of a programmer writing the commands to solve a problem, the program generates its own algorithm based on example data and a desired output. The machine-learning techniques that would later evolve into today\u2019s most powerful AI systems followed the latter path: the machine essentially programs itself.<\/p>\n\n\n\n At first this approach was of limited practical use, and in the 1960s and \u201970s it remained largely confined to the fringes of the field. Then the computerization of many industries and the emergence of large data sets renewed interest. That inspired the development of more powerful machine-learning techniques, especially new versions of one known as the artificial neural network. By the 1990s, neural networks could automatically digitize handwritten characters.<\/p>\n\n\n\n But it was not until the start of this decade, after several clever tweaks and refinements, that very large\u2014or \u201cdeep\u201d\u2014neural networks demonstrated dramatic improvements in automated perception. Deep learning is responsible for today\u2019s explosion of AI. It has given computers extraordinary powers, like the ability to recognize spoken words almost as well as a person could, a skill too complex to code into the machine by hand. Deep learning has transformed computer vision and dramatically improved machine translation. It is now being used to guide all sorts of key decisions in medicine, finance, manufacturing\u2014and beyond.<\/p>\n\n\n\n The workings of any machine-learning technology are inherently more opaque, even to computer scientists, than a hand-coded system. This is not to say that all future AI techniques will be equally unknowable. But by its nature, deep learning is a particularly dark black box.<\/p>\n\n\n\n You can\u2019t just look inside a deep neural network to see how it works. A network\u2019s reasoning is embedded in the behavior of thousands of simulated neurons, arranged into dozens or even hundreds of intricately interconnected layers. The neurons in the first layer each receive an input, like the intensity of a pixel in an image, and then perform a calculation before outputting a new signal. These outputs are fed, in a complex web, to the neurons in the next layer, and so on, until an overall output is produced. Plus, there is a process known as back-propagation that tweaks the calculations of individual neurons in a way that lets the network learn to produce a desired output.<\/p>\n\n\n\n The many layers in a deep network enable it to recognize things at different levels of abstraction. In a system designed to recognize dogs, for instance, the lower layers recognize simple things like outlines or color; higher layers recognize more complex stuff like fur or eyes; and the topmost layer identifies it all as a dog. The same approach can be applied, roughly speaking, to other inputs that lead a machine to teach itself: the sounds that make up words in speech, the letters and words that create sentences in text, or the steering-wheel movements required for driving.<\/p>\n\n\n\n Ingenious strategies have been used to try to capture and thus explain in more detail what\u2019s happening in such systems. In 2015, researchers at Google modified a deep-learning-based image recognition algorithm so that instead of spotting objects in photos, it would generate or modify them. By effectively running the algorithm in reverse, they could discover the features the program uses to recognize, say, a bird or building. The resulting images, produced by a project known as Deep Dream, showed grotesque, alien-like animals emerging from clouds and plants, and hallucinatory pagodas blooming across forests and mountain ranges. The images proved that deep learning need not be entirely inscrutable; they revealed that the algorithms home in on familiar visual features like a bird\u2019s beak or feathers. But the images also hinted at how different deep learning is from human perception, in that it might make something out of an artifact that we would know to ignore. Google researchers noted that when its algorithm generated images of a dumbbell, it also generated a human arm holding it. The machine had concluded that an arm was part of the thing.<\/p>\n\n\n\n Further progress has been made using ideas borrowed from neuroscience and cognitive science. A team led by Jeff Clune, an assistant professor at the University of Wyoming, has employed the AI equivalent of optical illusions to test deep neural networks. In 2015, Clune\u2019s group showed how certain images could fool such a network into perceiving things that aren\u2019t there, because the images exploit the low-level patterns the system searches for. One of Clune\u2019s collaborators, Jason Yosinski, also built a tool that acts like a probe stuck into a brain. His tool targets any neuron in the middle of the network and searches for the image that activates it the most. The images that turn up are abstract (imagine an impressionistic take on a flamingo or a school bus), highlighting the mysterious nature of the machine\u2019s perceptual abilities.<\/p>\n\n\n\n We need more than a glimpse of AI\u2019s thinking, however, and there is no easy solution. It is the interplay of calculations inside a deep neural network that is crucial to higher-level pattern recognition and complex decision-making, but those calculations are a quagmire of mathematical functions and variables. \u201cIf you had a very small neural network, you might be able to understand it,\u201d Jaakkola says. \u201cBut once it becomes very large, and it has thousands of units per layer and maybe hundreds of layers, then it becomes quite un-understandable.\u201d<\/p>\n\n\n\n In the office next to Jaakkola is Regina Barzilay, an MIT professor who is determined to apply machine learning to medicine. She was diagnosed with breast cancer a couple of years ago, at age 43. The diagnosis was shocking in itself, but Barzilay was also dismayed that cutting-edge statistical and machine-learning methods were not being used to help with oncological research or to guide patient treatment. She says AI has huge potential to revolutionize medicine, but realizing that potential will mean going beyond just medical records. She envisions using more of the raw data that she says is currently underutilized: \u201cimaging data, pathology data, all this information.\u201d<\/p>\n\n\n\n After she finished cancer treatment last year, Barzilay and her students began working with doctors at Massachusetts General Hospital to develop a system capable of mining pathology reports to identify patients with specific clinical characteristics that researchers might want to study. However, Barzilay understood that the system would need to explain its reasoning. So, together with Jaakkola and a student, she added a step: the system extracts and highlights snippets of text that are representative of a pattern it has discovered. Barzilay and her students are also developing a deep-learning algorithm capable of finding early signs of breast cancer in mammogram images, and they aim to give this system some ability to explain its reasoning, too. \u201cYou really need to have a loop where the machine and the human collaborate,\u201d -Barzilay says.<\/p>\n\n\n\n The U.S. military is pouring billions into projects that will use machine learning to pilot vehicles and aircraft, identify targets, and help analysts sift through huge piles of intelligence data. Here more than anywhere else, even more than in medicine, there is little room for algorithmic mystery, and the Department of Defense has identified explainability as a key stumbling block.<\/p>\n\n\n\n David Gunning, a program manager at the Defense Advanced Research Projects Agency, is overseeing the aptly named Explainable Artificial Intelligence program. A silver-haired veteran of the agency who previously oversaw the DARPA project that eventually led to the creation of Siri, Gunning says automation is creeping into countless areas of the military. Intelligence analysts are testing machine learning as a way of identifying patterns in vast amounts of surveillance data. Many autonomous ground vehicles and aircraft are being developed and tested. But soldiers probably won\u2019t feel comfortable in a robotic tank that doesn\u2019t explain itself to them, and analysts will be reluctant to act on information without some reasoning. \u201cIt\u2019s often the nature of these machine-learning systems that they produce a lot of false alarms, so an intel analyst really needs extra help to understand why a recommendation was made,\u201d Gunning says.<\/p>\n\n\n\n This March, DARPA chose 13 projects from academia and industry for funding under Gunning\u2019s program. Some of them could build on work led by Carlos Guestrin, a professor at the University of Washington. He and his colleagues have developed a way for machine-learning systems to provide a rationale for their outputs. Essentially, under this method a computer automatically finds a few examples from a data set and serves them up in a short explanation. A system designed to classify an e-mail message as coming from a terrorist, for example, might use many millions of messages in its training and decision-making. But using the Washington team\u2019s approach, it could highlight certain keywords found in a message. Guestrin\u2019s group has also devised ways for image recognition systems to hint at their reasoning by highlighting the parts of an image that were most significant.<\/p>\n\n\n\n One drawback to this approach and others like it, such as Barzilay\u2019s, is that the explanations provided will always be simplified, meaning some vital information may be lost along the way. \u201cWe haven\u2019t achieved the whole dream, which is where AI has a conversation with you, and it is able to explain,\u201d says Guestrin. \u201cWe\u2019re a long way from having truly interpretable AI.\u201d<\/p>\n\n\n\n It doesn\u2019t have to be a high-stakes situation like cancer diagnosis or military maneuvers for this to become an issue. Knowing AI\u2019s reasoning is also going to be crucial if the technology is to become a common and useful part of our daily lives. Tom Gruber, who leads the Siri team at Apple, says explainability is a key consideration for his team as it tries to make Siri a smarter and more capable virtual assistant. Gruber wouldn\u2019t discuss specific plans for Siri\u2019s future, but it\u2019s easy to imagine that if you receive a restaurant recommendation from Siri, you\u2019ll want to know what the reasoning was. Ruslan Salakhutdinov, director of AI research at Apple and an associate professor at Carnegie Mellon University, sees explainability as the core of the evolving relationship between humans and intelligent machines. \u201cIt\u2019s going to introduce trust,\u201d he says.<\/p>\n\n\n\n Just as many aspects of human behavior are impossible to explain in detail, perhaps it won\u2019t be possible for AI to explain everything it does. \u201cEven if somebody can give you a reasonable-sounding explanation [for his or her actions], it probably is incomplete, and the same could very well be true for AI,\u201d says Clune, of the University of Wyoming. \u201cIt might just be part of the nature of intelligence that only part of it is exposed to rational explanation. Some of it is just instinctual, or subconscious, or inscrutable.\u201d<\/p>\n\n\n\n If that\u2019s so, then at some stage we may have to simply trust AI\u2019s judgment or do without using it. Likewise, that judgment will have to incorporate social intelligence. Just as society is built upon a contract of expected behavior, we will need to design AI systems to respect and fit with our social norms. If we are to create robot tanks and other killing machines, it is important that their decision-making be consistent with our ethical judgments.<\/p>\n\n\n\n To probe these metaphysical concepts, I went to Tufts University to meet with Daniel Dennett, a renowned philosopher and cognitive scientist who studies consciousness and the mind. A chapter of Dennett\u2019s latest book, From Bacteria to Bach and Back, an encyclopedic treatise on consciousness, suggests that a natural part of the evolution of intelligence itself is the creation of systems capable of performing tasks their creators do not know how to do. \u201cThe question is, what accommodations do we have to make to do this wisely\u2014what standards do we demand of them, and of ourselves?\u201d he tells me in his cluttered office on the university\u2019s idyllic campus.<\/p>\n\n\n\n He also has a word of warning about the quest for explainability. \u201cI think by all means if we\u2019re going to use these things and rely on them, then let\u2019s get as firm a grip on how and why they\u2019re giving us the answers as possible,\u201d he says. But since there may be no perfect answer, we should be as cautious of AI explanations as we are of each other\u2019s\u2014no matter how clever a machine seems. \u201cIf it can\u2019t do better than us at explaining what it\u2019s doing,\u201d he says, \u201cthen don\u2019t trust it.\u201d<\/p>\n<\/div><\/div>\n\n\n\n RC:<\/strong> Esattamente! Di questo avevamo parlato nella nostra prima intervista, se ben ricordo. Eccetera eccetera\u2026 <\/p>\n\n\n\n Ometto la parte restante in quanto non attinente all’argomento di cui trattiamo. O meglio… forse qualcosa potrebbe tornarci utile in seguito. Vedremo.<\/p>\n\n\n\n
eccoti con questa email il bandolo da cui nei prossimi mesi proveremo a dipanare la matassa proverbiale.
Prima di tutto, per\u00f2, dovremmo stabilire certe regole cos\u00ec da renderci la vita il pi\u00f9 possibile semplice quando saremo arrivati alle conclusioni che speriamo siano illuminanti (…)<\/p>\n\n\n\n\n
Oppure, pi\u00f9 recentemente (nel XIX secolo), a Innocenzo Manzetti. Tuttavia anche in questi casi c\u2019\u00e8 un \u201cma\u201d, non essendoci alcun intento artistico ma soltanto emulativo della natura. Che poi si potrebbe dire sia il sogno a monte del concetto stesso di scienza. Replicare artificialmente, attraverso l\u2019ingegno, ci\u00f2 che Dio o il Creatore a cui preferisci attribuire l\u2019inizio di tutto.
Insomma, caro Riccardo, mi pare d\u2019averti ammorbato abbastanza, adesso a te la parola.
Manfredi
PS: Ricordami di raccontarti della fotografa di Hong Kong nella prox email. Non pensar male\u2026 \u00e8 attinente! ![\"?\"](\"https:\/\/s.w.org\/images\/core\/emoji\/15.0.3\/svg\/1f609.svg\")
<\/p>\n\n\n\n
Stavo pensando che forse il punto d\u2019arrivo potrebbe essere l\u2019arte nell\u2019era post-umana. Cio\u00e8 l\u2019uomo da poco non esiste pi\u00f9 ma le \u201cmacchine\u201d hanno scoperto l\u2019importanza dell\u2019arte. Forse nel tentativo di ricreare per se le emozioni che provava l\u2019essere umano. Una specie di retroingegneria al contrario. Oppure, se l\u2019uomo realmente finisse per integrarsi totalmente con le \u201cmacchine\u201d, queste avrebbero un patrimonio genetico in parte umano\u2026<\/p>\n\n\n\n
Ti lascio le mie prime impressioni a caldo. L\u2019idea di fare il passo finale e togliere gli artisti dal palcoscenico dell\u2019arte mi pare, a questo punto, un passo necessario, doveroso. Non perch\u00e9 io ritenga che non vi sia pi\u00f9 futuro per l\u2019arte, ma perch\u00e9 gli artisti e i critici non possono continuare a restare chiusi in una torre d\u2019avorio. Se si pone la questione della fine del lavoro, o addirittura della fine dell\u2019uomo, \u00e8 doveroso porre con tutta la forza possibile la questione della fine dell\u2019artista. Cosi\u2019 come ci si muove verso il post-umano, ci si muove di necessita\u2019 verso il post-artista. Se l\u2019artista vuole competere con le macchine, deve potenziarsi. Tanto pi\u00f9 che i critici, da decenni, hanno decretato la fine dell\u2019arte, la fine dell\u2019opera d\u2019arte, intesa in modo tradizionale. E gli artisti si sono adattati alla situazione. Non si pu\u00f2 pensare di decretare la fine dell\u2019arte e continuare a giocare intorno all\u2019altare dell\u2019arte, nella veste di critici o di artisti. Questo gioco ha ridotto l\u2019arte ad operazione mediatica. Ecco, io vedo l\u2019idea di un\u2019esposizione d\u2019opere d\u2019arte senza artisti come qualcosa che va ben oltre la mera provocazione. Non deve essere l\u2019ennesima provocazione che lascia il tempo che trova. L\u2019artista si propone al mondo come l\u2019esemplare d\u2019uomo pi\u00f9 libero e creativo \u2013 e dunque insostituibile. In realt\u00e0, per quanto l\u2019artista cerchi di fuggire ai clich\u00e9 \u00e8 sempre prigioniero degli stessi. Oggi, se un artista volesse essere davvero originale, se volesse davvero cantare fuori dal coro, dovrebbe dipingere come Michelangelo. Ma nessuno lo fa. Ci\u00f2 che fa l\u2019artista odierno pu\u00f2 essere prodotto da una macchina. Non riprodotto, ma prodotto. L\u2019intelligenza artificiale \u00e8 gi\u00e0 capace di produrre un comportamento imprevedibile, ossia non programmato. A nulla vale l\u2019argomento che la macchina, a differenza dell\u2019uomo, non \u00e8 cosciente. Che si possa trattare in modo dicotomico la questione della coscienza (c\u2019\u00e8 \/ non c\u2019\u00e8) \u00e8 frutto di una visione dualistica, fortemente radicata nella cultura occidentale, che non trova necessariamente riscontro nella realt\u00e0. Cos\u00ec come molti sono oggi propensi a riconoscere che anche gli animali posseggono un certo grado di coscienza, in altre culture extraeuropee sono gi\u00e0 pronti a riconoscere che essa \u00e8 presente anche nelle macchine, quand\u2019anche a livello della coscienza di un insetto. \u00c8 rimasto famoso un dibattito negli anni novanta tra David Chalmers e Christof Koch in cui si dibatteva la realt\u00e0 della coscienza. Il primo sosteneva che non \u00e8 insensato postulare che l\u2019informazione abbia uno status ontologico, come la materia e l\u2019energia. Il secondo negava questa posizione e finiva chiedendo a Chalmers: \u201cCome faccio a sapere che lei \u00e8 cosciente?\u201d. Il senso della domanda era che noi assumiamo che gli altri abbiano le stesse esperienze soggettive che noi abbiamo, ma non ne abbiamo certezza. Commentando il dialogo, il giornalista scientifico John Horgan scrisse: \u201cAfter all, one can certainly imagine a world of androids that resemble humans in every respect- except that they do not have conscious experience of the world<\/em>.\u201d In fondo potremmo gi\u00e0 essere in questo mondo di androidi, in una sorta di Matrix, senza saperlo. Possiamo scacciare lo spettro del solipsismo soltanto attraverso un postulato: che chi ci circonda \u00e8 dotato di coscienza. Nulla per\u00f2 vieta di fare la stessa assunzione a riguardo delle macchine pi\u00f9 sofisticate. Quello che \u00e8 un comportamento random, imprevedibile, potrebbe in realt\u00e0 essere un comportamento libero. Andr\u00f2 pi\u00f9 nei dettagli della questione nei prossimi messaggi.<\/p>\n\n\n\n\n
\n
\n
\n
\n
. . .<\/strong><\/li>\n<\/ol>\n\n\n\n
THE DARK SECRET AT THE HEART OF AI<\/strong>
No one really knows how the most advanced algorithms do what they do. That could be a problem.<\/em><\/p>\n\n\n\n
April 11, 2017<\/p>\n\n\n\n
Se ragioniamo in termini di \u201cpossibilita\u2019\u201d il discorso regge. Ora, come accenno nella risposta alla #6, il problema e\u2019 trovare le macchine adeguate per la giustificazione filosofica della scelta siamo sufficientemente attrezzati.<\/p>\n\n\n\n
\n\n\n\n