Man vs Magni: Nista Kompjuters Cook, Jiktbu u Paint Better Than Us?

Man v Machine: Can Computers Cook, Write and Paint Better Than Us?

intelliġenza artifiċjali issa jistgħu jirbħu logħba, jirrikonoxxu wiċċ tiegħek, anki tappella kontra biljett ipparkjar tiegħek. Iżda jista 'jagħmel l-għalf anke bnedmin isibu delikata?


Powered by Guardian.co.ukDan l-artiklu intitolat “Man vs magna: jista kompjuters kok, jiktbu u żebgħa aħjar minna?” ġie miktub minn Leo Benedictus, għal The Guardian nhar is-Sibt 4 Ġunju 2016 08.00 UTC

wieħed video, Għalija, mibdula kollox. Hu l filmati mill-logħba Atari qodma Aħrab, dak fejn inti slide jaqdfu xellug u tal-lemin tul il-qiegħ tal-iskrin, tipprova teqred briks mill bouncing ballun għal ġo fihom. Inti jista 'jkollok taqra dwar l-plejer tal-logħba: algoritmu żviluppat mill DeepMind, il-kumpanija Brittanika intelliġenza artifiċjali li AlphaGo programm wkoll tħabbat wieħed mill-akbar li qatt Mur plejers, Lee Sedol, aktar kmieni din is-sena.

Forsi inti tistenna kompjuter li tkun tajba fil-logħob tal-kompjuter? Ladarba dawn jafu x'għandhom jagħmlu, dawn żgur jagħmlu dan malajr u b'mod aktar konsistenti minn kull bniedem. Tbegħid plejer DeepMind tal naf xejn, madankollu. Ma kien ipprogrammat istruzzjonijiet dwar kif jaħdem il-logħba; din lanqas ma kienet qal kif jużaw il-kontrolli. Kollha li kellha kienet l-immaġini fuq l-iskrin u l-kmand biex nipprova nikseb punti kemm jista 'jkun.

Watch -video. Għall-ewwel, l-jaqdfu tikri l-waqgħa ballun fis oblivion, jafu ebda aħjar. eventwalment, biss mucking madwar, huwa knocks l-ballun lura, jeqred brick u gets punt, għalhekk hija tirrikonoxxi dan u ma dan aktar spiss. Wara prattika sagħtejn ", jew madwar 300 logħob, sar serjament tajba, aħjar milli suppost jew I qatt se jkun. imbagħad, wara madwar 600 logħob, affarijiet nikseb Spooky. L-algoritmu jibda bl-għan li l-istess post, terġa u terġa, sabiex bejta permezz tal-briks fl-ispazju wara. ladarba hemm, bħal kull attur tbegħid jaf, -ballun se bounce madwar għal filwaqt, ġbir punti ħielsa. Huwa strateġija tajba li l-kompjuter ħarāet bil waħdu.

"Meta r-riċerkaturi tagħna raw dan, li attwalment ixxukkjat minnhom,"CEO DeepMind s, Demis Hassabis, qal udjenza waqt konferenza teknoloġija f'Pariġi. Tista watch dimostrazzjoni tiegħu, wisq, u tisma 'l-Rires u applause meta l-magna figuri barra istrateġija tgħawwir tagħha. Il-kompjuter sar intelliġenti, daqsxejn bħalna.

"Intelliġenza artifiċjali" huwa biss dwar l-eqdem u l-aktar hyped ta 'frażijiet buzz kompjuters kollha tal-. L-idea kienet l-ewwel mooted serjament minn Alan Turing fil Computing Makkinarju u intelligence, l- 1950 karta li fiha pproponiet dak li sar magħruf bħala -test Turing: jekk magna tista 'tikkonvinċi lilek permezz konverżazzjoni li kien bniedem, dan kien isir kemm kull bniedem seta 'biex jipprova dan kien verament taħseb. Iżda l-AI terminu ma kienx ġeneralment użat sal- 1955, meta American matematiku John McCarthy pproponiet konferenza għal esperti. Din seħħet is-sena ta 'wara, u minn dakinhar l-għalqa jimxu fuq madwar ċiklu żewġ deċennju ta 'manija u disprament. (Riċerkaturi anke jkollhom terminu ġdid - "xitwa AI" - biex jiddeskrivu jispeċifika tagħha barra tal-moda. -1970 u 1990 kienu partikolarment ħarxa.)

Illum hemm manija ġdid, li jistenna differenti mill-oħrajn: taqbel fil-but tiegħek. A telefon tista 'taħbit il-ċampjin dinji taċ-ċess, jirrikonoxxu kanzunetti fuq ir-radju u l-istampi tat-tfal tiegħek, u tittraduċi vuċi tiegħek f'lingwa oħra. Il-robot Nao isaffru hawn Yotam Ottolenghi tista 'timxi fuq żewġ saqajn, jitkellmu, isibu ballun u anke żfin. (Huwa robot, għalkemm, mhux AI: hija ma jista 'disinn menu.)

Smigħ dwar l-avvanzi fit-AI, inti m'għandekx bżonn espert biex jgħidlek li tkun eċċitati, jew jibża. Inti biss tibda tikseb l-sentiment: intelliġenza hija hawnhekk. B'mod ċar Google ltqajna l-sentiment, wisq, minħabba li xtraw DeepMind għal $ 650 miljun rumored. Fl 2013, Facebook nediet proġett tagħha stess, mal-pjanijiet biex jiżviluppaw għarfien tal-lingwa tal-wiċċ u naturali għas-sit. Iżviluppaturi diġà bdew ix-xogħol fuq chatbots intelliġenti, li utenti Facebook se jkunu jistgħu jsejħu użu tas-servizz Messenger tiegħu.

S'issa, kompjuters ma kinux "intelliġenti" fil-livelli kollha, jew biss b'mod restrittiv hekk. Li ħadthom kien tajjeb fil-kompiti faċli li tgħammix magħna, bħal matematika, iżda ħażin lejn dawk nieħdu għall mogħtija, li tirriżulta li tkun serjament iebes. L-att ta 'mixi hija xi ħaġa robots moderna jitgħallmu bħal trabi u għadhom jissieltu ma'; ħidmiet Pöttering bażiċi jibqgħu ħolm mbiegħda. "Eżempju wieħed huwa l-faċilità li biha inti jew I tista 'tagħmel kikkra tè fil-kċina xi ħadd ieħor,"tgħid Professur Alan Winfield, a roboticist fl-Università ta 'l-Punent tal-Ingilterra. "Ma jkunx hemm robot fuq il-pjaneta li jistgħu jagħmlu dan."

Biex tifhem għaliex qed bniedem hija tant diffiċli, jaħsbu dwar kif inti tista 'tikseb kompjuter biex jirrikonoxxu nies minn ritratti. mingħajr AI, inti għandek tkun taf kif għandek tagħmel it yourself ewwel, sabiex programm tal-kompjuter. Inti għandek biex jiġbru u jaħsbu dwar il-mudelli possibbli, kuluri u forom ta 'uċuħ, u kif dawn jinbidlu fid-dawl u f'angoli differenti - u inti għandek tkun taf liema huwa sinifikanti u dak li huwa biss tajn fuq il-lenti. bl-AI, inti ma għandekx biex jispjegaw: inti biss jagħtu muntanji ta 'data reali għal kompjuter u ħalliha jitgħallmu. Kif inti disinn tas-software tat-tagħlim għadha kwistjoni esoteric, il-provinċja ta 'xi xjenzjati tal-kompjuter ftit mfittxija, iżda huwa ċar li ħadthom qbilna fuq li r-rebbieħ billi toħloq strutturi ta 'ipproċessar tad-data bbażata laxk fuq l-istrutturi fil-moħħ. (Din tissejjaħ "tagħlim fil-fond".) Fir-rigward tal-muntanji ta 'data reali, ukoll, dan huwa dak Google, facebook, Amazon, Über u l-bqija jiġri li tinsab madwar.

F'dan l-istadju, aħna għadhom ma jafux liema użi ta 'l-AI se jispiċċaw aħjar. Josh Newlan, Coder California jaħdmu f'Shanghai, ltqajna bored mal jisimgħu telefonati f'konferenza bla tmiem, hekk huwa mibni xi softwer li tisma għalih. Issa, kull meta l-isem Newlan huwa msemmi, kompjuter tiegħu istantanjament jibgħat lilu traskrizzjoni ta l-aħħar nofs minuta, tistenna 15 sekondi, imbagħad jilgħab reġistrazzjoni ta lilu qal, "Jiddispjacini, I ma jirrealizzaw mikrofonu tiegħi kienet fuq mutu. "Is-sena, Josh Browder, żagħżugħ Brittaniku, bniet avukat artifiċjali ħielsa li l-appelli kontra biljetti għall-parkeġġ; huwa pjanijiet biex jibnu ieħor biex jiggwidaw refuġjati permezz ta 'sistemi legali barranin. Il-possibilitajiet huma ... Ukoll, forsi algoritmu jista 'joqgħod il-possibbiltajiet.

Allura se imħuħ tal-magni xi darba jaqbeż iż tagħna stess? Ir-riċerkaturi nitkellem biex ikunu kawti, u jieħdu uġigħ biex jenfasizzaw dak magni tagħhom ma tistax tagħmel. Imma I iddeċieda li jpoġġi AI għat-test: jista 'jippjana ikla kif ukoll Ottolenghi? Jista 'żebgħa ritratt tiegħi? Hija t-teknoloġija għadha artifiċjalment intelliġenti - jew hija jibdew jiġu intelliġenti, ta 'vera?

It-test tat-tisjir

Well, I se ngħid mhuwiex horrible. Bnedmin servew lili agħar. Għalkemm fil-verità l-isem li Chef Watson IBM jagħti dan il-platt ("Zalza Chicken Fwied Savoury") huwa dwar kif appetizing kif jixraqlu.

Biex ikunu ġusti biex Chef Watson, u biex Guardian Weekend stess chef columnist Yotam Ottolenghi, Kelli jistabbilixxu lilhom pjuttost kompitu. I talab għal dixx bbażata fuq erba 'ingredjenti li deher li jappartjenu mkien viċin xulxin: fwied tat-tiġieġ, jogurt Grieg, wasabi u tequila. Huma jistgħu jżidu kwalunkwe ħaġa oħra li xtaqu, iżda dawk erbgħa kellhom ikunu fil-dixx lest, li nixtieq issajjar u jieklu. Chef Watson ma toqgħodx lura, istantanjament tagħti me żewġ zlazi għaġin. Ottolenghi kien aktar prudenti. "Meta sirt l-isfida ħsibt, "Dan mhux sejjer jaħdem,"" Hu jgħidlek me.

Ħsibt l-istess. Jew għall-inqas ħsibt se jispiċċaw jieklu żewġ platti li rnexxielhom tkun OK minkejja l-ingredjenti tagħhom, minflok minħabba minnhom. Fil-fatt - u inti ser taħseb me creep, imma iva, liema - riċetta Ottolenghi kienet rivelazzjoni: fwied u basal u tnaqqis tequila, notifikata bid tuffieħa, ravanell, pitravi u ċikwejra slaw, bil-wasabi u jogurt dressing. Il-dixx jista 'jagħmel ftit sens fuq il-karta, imma jien devoured sensazzjoni plateful li kull element ikkontrollata. (U vinaigrette magħqud mal-jogurt u wasabi minflok mustarda: serjament, jipprova hu.) Ottolenghi jgħidlekx me-riċetta hija biss qasir whisker ta pubblikabbli.

Il-ħaġa hija, li dixx ħadet lilu u tiegħu t-tim tlett ijiem biex jiġu perfezzjonati. Huma kienu kapaċi biex iduqu u jiddiskutu togħmiet, nisġa, kuluri, temperaturi, b'mod li Watson ma tistax - għalkemm kien hemm "diskussjonijiet" dwar żżid mekkaniżmu feedback fil-futur, Chef inġinier ċomb Watson, Florian Pinel, jgħidlekx me. "A riċetta hija tali ħaġa kumplessa,"Ottolenghi tgħid. "Huwa diffiċli għalija anke li wieħed jifhem kif il-kompjuter ser javviċina dan."

Yotam Ottolenghi u platti Chef Watson
Yotam Ottolenghi u platti Chef Watson Ritratt: Jay Brooks għall-Gwardjan

Watson ewwel nbniet minn IBM biex tirbaħ il-periklu gameshow televiżjoni! fil 2011. F'xi metodi kienet sfida qarrieqa, minħabba li għal kompjuter l-parti iebsa minn kwizz huwa fehim-mistoqsijiet, ma kienux jafu l-tweġibiet; għall-bnedmin, huwa l-mod ieħor madwar. Iżda Watson rebaħ, u t-teknoloġija tagħha bdiet tiġi applikata bnadi oħra, inkluż bħala chef, jiġġeneraw riċetti ġodda bbażati fuq 10,000 Eżempji reali meħuda mill rivista bon appétit.

L-ewwel l-software kellhom "jixrob" dawn riċetti, kif it-tim Watson poġġih. A lott ta 'komputazzjoni marru fil-fehim dak l-ingredjenti kinux, kif dawn ġew ippreparati, kemm dawn kienu imsajjar għall, sabiex ikunu jistgħu jispjegaw kif jużawhom fil platti ġodda. (Il-proċess għadu jista 'jmur awry. Anke issa Chef Watson jirrakkomanda ingredjent imsejjaħ "Mollusk", li siewi jispjega huwa "s-sitt full-tul album billi Ween".)

A problema akbar kien qed jipprova jagħti l-magna sens ta 'togħma. "Huwa faċli biżżejjed għal kompjuter biex joħolqu f'kombinazzjoni novella,"Pinel tgħid, "Imma kif jista jevalwa waħda?"Watson kien mgħallma biex jikkunsidraw kull ingredjent bħala taħlita ta 'komposti tal-benna li speëifiëi - li minnhom hemm eluf - u mbagħad biex jikkombinaw l-ingredjenti li kellhom komposti komuni. (Dan il-prinċipju, tqabbil ikel, hija stabbilita sewwa fost il-bnedmin.) Fl-aħħarnett, -software jiġġenera pass pass istruzzjonijiet li jagħmlu sens għal kok bniedem. L-enfasi hija fuq sorpriżi aktar milli l-ippjanar prattiku ikla. "Chef Watson huwa verament hemm biex jispiraw inti,"Pinel jispjega. Kull riċetta jiġi ma 'l tfakkira li "jużaw il-kreattività tiegħek u ġudizzju".

U għandi bżonn biex. L-ewwel pass huwa li "toast ċatt weraq tursin", li biss mhix idea tajba. I am jagħmlu, effettiv, a majjal imħawra bil-mod imsajjar u ragu ċanga, inklużi erba ingredjenti kollha tiegħi, iżda Watson oddly jinkludi wkoll ħjar u jżomm me javżak li "staġun mal allspice", li Jien nirrifjuta li jagħmlu fuq il-prinċipju. Fl-aħħar, I jkollhom zalza sinjuri b'togħma pjuttost qrib il-irziezet, iżda mhux uneatable. I ma tista 'togħma l wasabi jew l tequila, li Ninsab kuntenta dwar.

Yotam Ottolenghi ma robot Nao
Yotam Ottolenghi ma robot Nao mislufa korteżija ta 'skola primarja Heber, London. Ritratt: Jay Brooks. Styling: Lee Flude

Watson huwa għaqlija u l-kompitu hija iebsa, imma jiena lest li jgħidu li dan ma jkunx aktar minn daqsxejn ta 'gost għall nerds ikel, sakemm Ottolenghi jieqaf lili. "Naħseb li l-idea ta 'bil-mod tisjir l-fwied bi ftit ta' laħam hija kbira," hu qal. "Hija jintensifika l-togħma. Kollox se jingħaqdu flimkien. Jekk I kellha tibda mill-ġdid ma 'dan riċetta, ovvjament il jogurt ma jinstallax - iżda nixtieq jħallu l-ġilda oranġjo hemm, ftit mill-ħwawar. Ma naħsibx li huwa riċetta ħażina ħafna. Hija tista 'taħdem. "

Verdett Watson ġlud weirdness ta 'l-ingredjenti, iżda Ottolenghi jagħmilhom nijet.

It-test bil-miktub

Poġġi ftit wordsmith li jmiss għall-magni fearsome ta 'IBM u Google, u jidher li computationally avvanzati bħala b'kalkulatur. Iżda filwaqt Watson fumbles permezz ta 'apprendistat tagħha, Wordsmith diġà fuq ix-xogħol. If you read stock market reports from the Associated Press, or Yahoo’s sports journalism, there is a good chance you’ll think they were written by a person.

Wordsmith is an artificial writer. Developed by a company in North Carolina called Automated Insights, it plucks the most interesting nuggets from a dataset and uses them to structure an article (or email, or product listing). When it comes across really big news, it uses more emotive language. It varies diction and syntax to make its work more readable. Even a clumsy robot chef can have its uses, but writing for human readers must be smooth. Hooked up to a voice-recognition device such as Amazon’s Echo, Wordsmith can even respond to a spoken human question – about the performance of one’s investments, say – with a thoughtfully spoken answer, announcing what’s interesting first, and leaving out what isn’t interesting at all. If you didn’t know the trick, you’d think Hal 9000 had arrived.

The trick is this: Wordsmith does the part of writing that people don’t realise is easy. Locky Stewart from Automated Insights gives me a tutorial. You write into Wordsmith a sentence such as, “New ABC figures show that the New York Inquirer’s circulation rose 3% in April.” Then you play around. Il 3% has come from your data, so you select the word “rose” and write a rule, known as a “branch”, which will change the word “rose” to the phrase “shot up” if the percentage is more than 5%. Then you branch “rose” to become “fell” if the percentage is negative. If the percentage is -5% or lower, “rose” becomes “plummeted”.

Then you feed it synonyms. So “plummeted” can also be “fell sharply by”. “The Inquirer’s circulation” can be “circulation at the Inquirer”. “Shot up” can be “soared” and so on. Then you add more sentences, perhaps about online traffic, or about which days’ print copies sold best, or about comparisons year-on-year. Then you get clever. You tell Wordsmith to put the sentences with the most newsworthy information first, defined perhaps as those that feature the greatest percentage changes. Maybe you add a branch to say that a result is “the best/worst performance among the quality titles”. Hell, you can even teach it some old Fleet Street tricks, so that if circulation plummets the piece begins “Editor Charles Kane is facing fierce criticism as”, but if circulation has “shot up” this becomes “Charles Kane has silenced critics with news that”. Insert “more” or “again” or “continues” if you get the same thing two months in a row.

“The artificial intelligence is actually the human intelligence that is building the network of logic,” Stewart says, “the same network you would use when writing a story. It could have been developed 10 jew 15 snin ilu, in code, but to make it work at this scale has only been possible lately.” Clearly it takes longer to prepare an article on Wordsmith than to write one conventionally, but once you’ve done so, the computer can publish a fresh newspaper circulation story every month, on every newspaper, within seconds of receiving the information. It can publish millions of stories in minutes – or publish only some of them, if the data doesn’t reach a given threshold of newsworthiness. Thus it becomes an automated editor, wisq, with adjustable tastes in thoroughness, frequency and hysteria.

For Wordsmith’s task, I suggest football: it’s a field that produces a lot of data and has a readership that wants personalised articles. Guardian football writer Jacob Steinberg volunteers to take on the computer, and I provide a table of facts from the recent Premier League: last season’s league position and this season’s position at Christmas and at the end, goals scored and conceded, top scorer’s name and total, value of summer transfers and a quote from the manager.

Working solely from this data, computer and human must each write a review of the season for a given club. Steinberg chooses Leicester City on the basis that its numbers should contain a story that anyone would see. Wordsmith doesn’t need to choose. It will do all 20.

And in fact both computer and human quickly produce quite similar work:

Leicester City footballer Jamie Vardy

Both Steinberg and Wordsmith deliver dramatic first sentences. Perhaps keen to sound authentic, Automated Insights use some clever tricks to put feeling into the latter’s article, astutely guessing that Leicester were “hoping to finish in the top 10 after a 14th place finish last season”. I look through Wordsmith’s other articles and Southampton, having finished seventh last season, have “eyes on a European spot”, while Manchester City “began the season dreaming of a league title after finishing second”.

Conversely, Steinberg digs more meaningfully into the numbers, showing that Jamie Vardy not only scored 24 goals, but that this was a higher percentage of his team’s goals than was managed by all but two other players. Knowing how Wordsmith works, tal-kors, one could easily set it up to do the same. In fact looking through it, Steinberg’s entire article could have been created by a skilled Wordsmith programmer – with the exception of one line. “It’s a magical season,” he quotes the Leicester manager as saying, before adding, “justifiably so, given that a summer expenditure of £26.7m on transfers made them the eighth lowest spenders”. That “justifiably so” shows a writer who actually understands what he is writing.

Verdett Steinberg is a much better writer, unless you want 20 data-heavy articles in 10 minuti.

The painting test

A laptop wants me to smile. “It’s in a good mood," Simon Colton jgħid. He knows because he’s the scientist who programmed it. We are in the Science Museum in London, where the Painting Fool, as it is called, is giving a public demonstration. It’s important that I don’t show my teeth, Colton says, because something about the light makes them look green to the Painting Fool.

From my toothless smile the laptop creates a “conception” of what it would like to paint, based on its mood. The mood comes from a “sentiment analysis” of recent Guardian articles, as it happens (on average reading the Guardian is a downer, apparentement, apart from the stuff about gardening). Yesterday the Fool was in such a bad mood that it sent someone away unpainted; today it is feeling “positive”.

Next the Fool attempts to paint with a simulated brush and a simulated hand (actually, an image of Colton’s hand) on the screen behind me. It learned to reflect its mood from the work of Dan Ventura, another computer scientist, at Brigham Young University in Utah, who trained a neural network to recognise the emotional attributes of images by sitting thousands of people in front of tens of thousands of paintings and asking them to tag each one with whatever adjectives came to mind. The Fool now knows that bright colours reflect a good mood, and “pencils with tight hatching” create a picture that is “cold”. When it is done, it prints out a page with a typed self-critique. “Overall, this is quite a bright portrait,” it says. “That’s OK, but my style has lowered the level of bright here. So I’m a bit annoyed about that.”

Here along with us, intrigued but too busy at her easel to watch, huwa Sarah Jane Moon, an artist who exhibits with the Royal Society of Portrait Painters. She doesn’t want to see my teeth, either. “We paint from life,"Tgħid, “and you can’t hold a smile for sitting upon sitting. That’s why all the traditional portraits show quite relaxed features.”

The Painting Fool is a special machine, and even slightly famous, but I can’t deny that Moon is almost all of why I’m excited to be here. The feeling of being painted by a real person, having them look at you and think about you, is exciting and flattering. Sentiment analysis and training data, minn naha l-ohra, don’t add up to anything whose view of me I care about, and the finished portraits do not change my mind. Moon’s is a lovely, real thing, which feels straight away like one person seen by another. The Fool’s three efforts have qualities I like, but mostly they look like photographs that have gone through some kind of software filter. Colton insists the Fool is here “to learn to be better” but I look and think: so what?

Painting of Leo Benedictus by Sarah Jane Moon
Leo Benedictus as seen by Sarah Jane Moon…
Painting of Leo Benedictus by the Painting Fool computer
…and as imagined by the Painting Fool laptop. Ritratt: Murray Ballard

Then I think some more. For one thing, it turns out that art is more mechanical than I’d realised. “I try to look at Leo as an abstract set of shapes, formoli, kuluri, tones,” Moon tells Colton, “to get away from the fact that that’s a nose. Because when you start to do that, you get caught up in what you think looks like a nose.”

“What the software does is break it down into colour regions,” Colton says.

"Iva, eżattament,” Moon agrees. “I think that’s what the best painters do. It’s transcribing.” Afterwards she tells me she felt a kind of “kinship” with the software as they worked side by side.

aktar importanti minn hekk, I realise that what matters isn’t how the machine paints; it’s how I see. Moon I understand, Naħseb. She’s a person and I know how that feels, so I care about her picture. But what does it feel like to be the Painting Fool? Is that what its portraits are trying to tell me?

Verdett Moon’s painting is far richer; the Fool is still learning and has centuries of practice to go.

The translation test

Google Translate was the first piece of proper science fiction to come true, u it’s already a decade old. In many ways it typifies where AI has got to. Useful, żgur; impressive, without question; but still clunky as hell, despite big improvements.

If you haven’t used it, it works like this: enter text or web links in any of 103 supported languages and you get a rough translation seconds later in any of the others. The app on your phone will transcribe what you say and then speak it back, tradotta (32 languages supported); it can replace the text of a foreign language sign or menu wherever you point the camera. No explanation is needed of how cool that is (and it’s free).

Globally, half a billion people use Google Translate each month, mostly those who don’t speak English (which is 80% of people) but who want to understand the internet (which is 50% Ingliż). “Most of our growth, and actually most of our traffic, comes from developing or emerging markets such as Brazil, Indoneżja, Indja, Tajlandja,” says Barak Turovsky, head of product management and user experience at Google Translate. It’s surprisingly popular for dating, wisq, he adds. “Things like ‘I love you’ and ‘You have beautiful eyes’, that’s very prevalent.”

The software has always used a form of statistical machine learning: scouring the internet for already translated text – UN declarations, EU documents – and mapping the likelihood of certain words and phrases corresponding to one another. The more data it gathers, the better it gets, but the improvement levelled off a couple of years ago. Ftit, Turovsky says, they will deploy new deep learning algorithms, which will produce much more fluent translations.

Anke jekk, there are limits, and some seem fundamental when you talk to a human translator and realise how subtle their work is. Ros Schwartz u Anne de Freyman volunteer for this task. Both are professional French/English translators, and I need two because, in order to judge how good the translation is without being fluent in both languages, we need to translate twice – once out of English into French, once back again. Google Translate keeps no memory of the original and can do the same thing.

I choose a short passage of distinctive but not especially wild or ambiguous prose from the beginning of Herzog by Saul Bellow. Translators normally require context, so I tell Schwartz and De Freyman that it comes from a famous mid-century American novel.

Within a few days, Schwartz and De Freyman return a very smooth facsimile of the original text. Here and there some nuances have not survived, but the passage remains a pleasure to read, and the main meanings come across exactly.

Google Translate takes only a few seconds, and the result is both impressive and inadequate, weirdly good in places, in others weirdly bad – turning “he” into “it” and concocting the idea that Herzog is in love. Miraculously, it keeps “cracked” as a description of the hero. French has no word that combines the sense of “broken” and “mad” that cracked coveys in English, so De Freyman makes it “cinglé”, which comes back from Schwartz as “crazy”.

“Google Translate would look at statistical probability and say, what does ‘cracked’ mean?” Turovsky explains. “And statistically, it will try to decide whether it means ‘cracked’ or ‘crazy’ or whatever. dik, for a machine, is a non-trivial task.” Nor is it simple for a human, even though we find it easy. You’d have to ask whether Bellow could have meant that Herzog was “cracked” as in physically fractured. Then you’d have to assume not, because human bodies don’t generally do that. So you’d wonder what he did mean and assume instead, if you were not already familiar with the usage, that he must mean “crazy”, because you understand the rest of what you’ve read. But to do all this, wouldn’t Google Translate have to be pretty much conscious, I ask? Turovsky laughs. “I don’t think I’m qualified to answer that question.”

Verdett Some bullseyes and howlers from Google Translate, while Schwartz and De Freyman are fluent and exact.

guardian.co.uk © Guardian News & Media Limited 2010

Artikli relatati