Contra "Grandmaster-level Chess Didi Manɔmee" (2024): Nusitae Kpɔɖeŋudzesi ɖeɖeko Dzudzɔ Kpuie

Google DeepMind ƒe ƒe 2024 ƒe agbalẽ si gblɔ be grandmaster-level chess si me didi ƒe mɔnu xoxowo mele o la he ɖikeke vɛ enumake eye woɖo kpe edzi nyuie le AI numekukuhabɔbɔa katã me. Contra nyaʋiʋliawo ɖe seɖoƒe veviwo fia le raw pattern recognition ɖɔliɖɔli ɖe systematic analysis teƒe — nusɔsrɔ̃ siwo keke ta yi ŋgɔ sasasã wu chess yi asitsatsa ƒe automation, nyametsotsowɔwɔ ƒe ɖoɖowo, kple alesi platforms abe Mewayz architect intelligent workflows na over 138,000 users.

Nuka tututue Agbalẽ Gbãtɔa Gblɔ?

| To vovo na mɔ̃wo abe Stockfish alo AlphaZero, siwo dzroa etsɔme nɔƒe akpe geɖe va ɖo miliɔn geɖe me hafi tiaa ʋuʋu la, mɔnu sia nɔ te ɖe ahɔhɔ̃mekawo ƒe kadodo si wɔa nyagblɔɖi siwo to mɔ ɖeka dzi — vevietɔ "intuiting" the best move from pattern recognition alone.

Nya si wogblɔ la nye dzideƒotɔe: ne kpɔɖeŋu aɖe ate ŋu axɔ nɔƒe gɔmesese si sɔ tso hehenana ŋuti nyatakakawo me la, akɔntabubu si me wozãa ŋusẽ vɔ̃ɖi le ate ŋu ava zu nusi mehiã o. Gbãtɔ ƒe dodokpɔ me tsonu dze abe ɖe ŋugbedodo le eme ene, eye kpɔɖeŋua ɖo Elo ƒe dzesidewo gbɔ le grandmaster ƒe domedome le dodokpɔ ƒe nɔnɔme tɔxɛwo me.

Nukatae Ðeklemiɖelawo ʋlia nya be womeɖe didi ɖa vavã gbeɖe o?

Contra nyaʋiʋli si ŋu kakaɖedzi le wu la tɔa ŋku agbalẽa ƒe gɔmeɖoanyi vevitɔ. Wona hehe transformer la le teƒe miliɔn geɖe siwo Stockfish — mɔ̃ si ɖoa ŋu ɖe didi deto ŋu vevie. Ðeklemiɖelawo ʋlia nya be kpɔɖeŋua meɖe didi ɖa o; e ɖea tsie. Ðeko wotsɔ didia de ŋgɔe ɖe hehenana ƒe nyatakakawo me tsɔ wu be woawɔe le nutsotsoɣi.

ƒe nyawo

"Ne ègblɔ be kpɔɖeŋu aɖe ƒoa chess 'didi manɔmee' esime wòle hehe name le mɔ̃ si wotu ɖe didi dzi ƒe emetsonuwo ŋu le abe gbɔgblɔ be yekpɔ maze aɖe gbɔ anyigbatata manɔmee ene — le egbɔkpɔnu si ame bubu kpɔ to anyigbatata zazã me léle ɖe susu me vɔ megbe."

ƒe nyawo

Vovototo sia le vevie ŋutɔ. Kpɔɖeŋua srɔ̃ numekuku me tsonu ƒe nɔnɔmetata siwo womiã, ke menye ɖoƒe gɔmesese si le eɖokui si o. Ðe hehenana ƒe dzesi si wokpɔ tso didi me ɖa, eye dɔwɔwɔ adze anyi. Esia sɔ kple wo nɔewo tẽ le asitsatsa ŋuti nunya me: nyametsotsodɔwɔnu ɖesiaɖe si wotu ɖe AI dzi la nyo abe numekuku si wowɔ ɖe ɖoɖo nu si wotsɔ de eƒe hehenana ƒe mɔ̃a me ene ko.

Afikae Kpɔɖeŋu Dzadzɛnyenye Dzesidede Gblẽna le Dɔwɔwɔ Me?

Dodokpɔ si numekula siwo le wo ɖokui si wɔ la ɖe kpododonu ƒe nɔnɔme vevi siwo dzidzenu gbãtɔwo do viviti ɖe go:

ƒe nyawo
• Aɖaŋuɖoɖo ƒe nɔƒe goglowo: Kpɔɖeŋua to ƒokpli siwo hiã akɔntabubu wu ʋuʋu 4-5 ŋu ɣesiaɣi, afisi mɔ̃ xoxowo nyo wu to diditi siwo dze ƒã me.
• Fefe ƒe nuwuwu ƒe nɔnɔme yeyewo: Ðoƒe siwo le hehenana ƒe mama godo ɖe kpɔɖeŋua ƒe ŋutete be wòabu tame tso gɔmeɖose gbãtɔwo dzi la ɖe go, si he vodada vevi siwo amegbetɔ tɔgbuitɔ aɖeke mawɔ vɛ.
• Adversarial robustness: Ne tsitretsiɖeŋulawo ɖoe koŋ kplɔ fefewo yi teƒe siwo mebɔ o la, model la ƒe Elo ɖiɖi ŋutɔ — si fia be woalé nu ɖe susu me tsɔ wu be woase egɔme vavã.
• Nuwɔwɔ ɖekae le nyaƒoɖeamenu me: Togbɔ be dɔwɔwɔ le mama dedie nu dze abe grandmaster ƒe ɖoƒe ene hã la, vovototoa lolo sã wu amegbetɔ ƒe grandmasters alo search-based engines, eye afɔku gãwo dzɔna le agbɔsɔsɔ si mewɔ ɖeka kple grandmaster fefe vavãtɔ o.
• Positional complexity scaling: Esi board complexity dzi ɖe edzi la, dometsotso si le search-free model kple search-based engines dome la keke ta ɖe enu ŋutɔ tsɔ wu be wòanye linearly.

ƒe nyawo

Nukae Nyaʋiʋli Sia Gblɔ na AI-Dziɖuɖu ƒe Dɔwɔɖoɖowo?

Chess-didi manɔmee ƒe nyaʋiʋlia klẽ ɖe masɔmasɔ aɖe si le egbegbe AI zazã ƒe dzime dzi. Kpɔɖeŋudzesi kple ɖoɖowɔɖi me dzodzro menye nusiwo woate ŋu atrɔ o — wo kpena ɖe wo nɔewo ŋu. Nuɖoanyi siwo wɔa dɔ wu la ƒoa ŋuɖoɖo siwo me kɔ kabakaba kple susuŋudɔwɔwɔ si woɖo ɖe ɖoɖo nu nu ƒu le afisiwo afɔkuwo lolo le.

Esia tututue nye xɔtuɖaŋu si le megbe na Mewayz ƒe 207-module business operating system. Le esi teƒe be wòaɖo ŋu ɖe kpɔɖeŋu-match heuristics alo sedziwɔwɔ dzro ko ŋu la, mɔ̃a ƒoa mɔnu eveawo nu ƒu ɖekae le eƒe dɔwɔwɔ ƒe nuwo wɔwɔ le wo ɖokui si, CRM, dɔa dzikpɔkpɔ, kple ganyawo ƒe modules me. Aɖaŋuɖoɖo siwo wotu ɖe kpɔɖeŋu dzi kabakaba kpɔa nyametsotso siwo wowɔna edziedzi gbɔ, esime numekuku ƒe ɖoɖo siwo woɖo ɖe ɖoɖo nu ƒoa wo ɖokui ɖe nɔnɔme sesẽwo me — si nye alesi chess mɔ̃ sesẽtɔwo tsɔa ahɔhɔ̃mekawo ƒe dɔwɔwɔ ƒe dodokpɔ kple didi si woɖo taɖodzinu na doa ƒomee ƒe ahuhɔ̃e.

Nusɔsrɔ̃ si tso contra numekukua me me kɔ: ɖoɖo siwo gblɔna be yewoɖea nuŋububu si wowɔ ɖe ɖoɖo nu ɖa tsɔ doa alɔ nukpɔkpɔ dzadzɛ la ƒoa dɔwɔwɔ ƒe dzidzenuwo godoo. Eɖanye chess ƒe ɖoƒe aɖe dzi kpɔkpɔ alo asitsatsa ƒe mɔ̃ aɖe dzi kpɔkpɔ o, kpɔɖeŋu ƒe dzesidede kabakaba kple numekuku si woɖo koŋ ƒo ƒu wɔa dɔ wu mɔnu eveawo dometɔ ɖesiaɖe ɣesiaɣi le wo ɖokui si.

Aleke Wòle Be Míada "Ŋgɔyiyi" AI ƒe Nyawo Me Le Yi Ŋgɔ?

Contra nyaʋiʋliawo ɖoa ɖoɖo nyui aɖe si dzi woato ada AI numekuku ƒe nya siwo wodi vevie la kpɔ vevie. Gbã la, lé ŋku ɖe eŋu nenye be wokpɔ ŋutete si wogblɔ la gbɔ vavã alo ɖeko wogbugbɔe mama — ɖe ɖoɖoa ɖe didi ɖa, alo ɣlae ɖe hehenana ƒe ɖoɖoa mea? Evelia, dodokpɔ ƒe dɔwɔwɔ le nusiwo wotsɔ tsi tre ɖe wo nɔewo ŋu kple esiwo mele mama me o ŋu, ke menye dzidzenu nyuiwo ɖeɖeko o. Etɔ̃lia, dzidze nusiwo mewɔa tɔtrɔ o kple dɔwɔwɔ le nɔnɔme vɔ̃ɖitɔ kekeake me kpe ɖe mamã dediewo ŋu, elabena ɖoɖo si ƒoa fefe nyuie le ɣeyiɣi 90% me gake wòwɔa vodada gã aɖe le ɣeyiɣi 10% me la menye grandmaster-level le gɔmesese aɖeke nu o.

Dzadzraɖo ƒe gɔmeɖose siawo wɔa dɔ sɔsɔe ne asitsahawo le dɔwɔnu siwo ŋu AI-ŋusẽ le la me dzrom na woƒe dɔwɔnawo. Anyigbadzidzidzenuwo ateŋu atsyɔ gbɔdzɔgbɔdzɔ vevi siwo dona le xexeame ƒe nɔnɔme ŋutɔŋutɔwo me — nu ŋutɔŋutɔ si na nyanya Mewayz ƒe mɔnu si wòzã tsɔ tu kakaɖedzi ɖo le eƒe module ecosystem bliboa me.

Nyabiase Siwo Wobiana Enuenu

Ðe chess-without-search ƒe kpɔɖeŋua ɖo grandmaster ƒe ɖoƒe ŋutɔŋutɔa?

Le benchmark nɔnɔme siwo dzi wokpɔna me la, model la ɖo Elo dzidzedzekpɔkpɔwo gbɔ le grandmaster ƒe domedome. Gake dodokpɔ siwo wowɔ le wo ɖokui si ɖe masɔmasɔwo, tsitretsiɖeŋu ƒe gbɔdzɔgbɔdzɔwo, kple aɖaŋudzedze ƒe ŋkumaʋumaʋu goglo siwo gblẽa grandmaster ƒe hatsotsowo me fia. Grandmaster fefe vavãtɔ bia kakaɖedzi kple goglome si kpɔɖeŋua meɖe fia ɣesiaɣi o, si wɔe be nya si wogblɔ la le kpuie le mɔ̃ɖaŋununya gome tsɔ wu be wòasɔ le mɔ gbadza nu.

Ðe AI chess numekuku si me didi mele o gakpɔtɔ xɔ asi togbɔ be ɖeklemiɖeɖe siawo li hã?

Kakaɖedzitɔe. Numekukua ɖee fia be mɔ̃ siwo trɔa asi le nu ŋu ƒe xɔtuɖaŋuwo ate ŋu aƒo chess ŋuti sidzedze gbogbo aɖewo nu ƒu wòazu dodokpɔ siwo wowɔna kabakaba le mɔ ɖeka dzi. Esia ƒe dɔwɔwɔ ŋutɔŋutɔ le numekuku siwo wowɔna kabakaba, hehenana ƒe kpekpeɖeŋu, kple ɖoɖo siwo wotsɔ tsakae me. Contra arguments meɖea numekukua ɖa o — wotsɔa eƒe seɖoƒewo ɖoa nya siwo ƒo xlãe me nyuie eye wotsɔa nyataƒoƒo si wogblɔ wògbɔ eme la tsia tre ɖe eŋu.

Aleke nyaʋiʋli sia do ƒome kple asitsatsa ƒe nuwo wɔwɔ le wo ɖokui si ƒe dɔwɔnuwo tiatia?

Nusɔsrɔ̃ vevitɔe nye be nuwo wɔwɔ le wo ɖokui si nyuie bia be woatsɔ susuŋudɔwɔwɔ ƒe mɔnu nyuitɔ asɔ kple dɔ ƒomevi ɖesiaɖe. Nyametsotso bɔbɔe siwo wogbugbɔna wɔna la ɖea vi tso kpɔɖeŋu ƒe dzesidede kabakaba me. Nyametsotso sesẽ siwo me afɔku le ŋutɔ bia be woaku nu me le ɖoɖo nu. Mɔ̃ nyuitɔwo — abe Mewayz ƒe asitsatsa ƒe OS si wotsɔ wɔ ɖekae ene — ƒoa evea siaa nu ƒu, si kpɔa egbɔ be mɔnu ɖeka aɖeke mezua mɔxenu alo kpododonu ƒe teƒe le wò dɔwɔnawo katã me o.

Èle klalo be yeawɔ yeƒe dɔwɔƒea le ɖoɖo si wotu na duƒuƒu kple goglome siaa dzia? Mewayz ƒoa modules 207 siwo wotsɔ wɔ ɖekae kple automation si me nunya le si wowɔ na xexeame ŋutɔŋutɔ ƒe nu sesẽwo — ke menye benchmark theatrics o. Ðoɖowo dzea egɔme tso $19/mo na ƒuƒoƒo siwo bia kakaɖedzi le ɖoƒe ɖesiaɖe. Dze wò dodokpɔ femaxee gɔme le app.mewayz.com eye nàkpɔ alesi asitsadɔwɔɖoɖo vavãtɔ sena le eɖokui me.

katã me o