Ingabe i-AI ingabikezela imakethe yamasheya?

Isingeniso

Ukubikezela imakethe yamasheya sekuyisikhathi eside kuyinto "engcwele" efunwa ngabatshalizimali bezikhungo kanye nabathengisi emhlabeni jikelele. Ngentuthuko yakamuva ku- Artificial Intelligence (AI) kanye nokufunda komshini (ML) , abaningi bayazibuza ukuthi lobu buchwepheshe bugcine buvule imfihlo yokubikezela amanani esitoko. Ingabe i-AI ingabikezela imakethe yamasheya? Leli phepha elimhlophe lihlola lowo mbuzo ngombono womhlaba wonke, lichaza ukuthi amamodeli aqhutshwa yi-AI azama kanjani ukubikezela ukunyakaza kwemakethe, izisekelo zemfundiso ngemuva kwala mamodeli, kanye nemikhawulo yangempela abhekene nayo. Sethula ukuhlaziywa okungachemile, okusekelwe ocwaningweni kunokuba kube yi-hype, ngalokho i-AI engakwenza futhi engenakukwenza kumongo wokubikezela imakethe yezezimali.

Embonweni wezezimali, inselele yokubikezela igcizelelwa yi- Efficient Market Hypothesis (EMH) . I-EMH (ikakhulukazi ngesimo sayo "esiqinile") iphakamisa ukuthi amanani esitoko abonisa ngokugcwele lonke ulwazi olutholakalayo nganoma yisiphi isikhathi, okusho ukuthi akekho umtshali-zimali (ngisho nabantu bangaphakathi) ongakwazi ukudlula imakethe ngokuqhubekayo ngokuhweba ngolwazi olutholakalayo ( Amamodeli okubikezela amasheya aqhutshwa idatha asekelwe kumanethiwekhi e-neural: Ukubuyekezwa ). Ngamagama alula, uma izimakethe zisebenza kahle kakhulu futhi amanani ehamba ngendlela engahleliwe , khona-ke ukubikezela ngokunembile amanani esikhathi esizayo kufanele kube yinto engenakwenzeka. Naphezu kwalombono, ukuheha kokunqoba imakethe kuye kwakhuthaza ucwaningo olubanzi ezindleleni zokubikezela ezithuthukisiwe. I-AI nokufunda komshini sekuyinto ebalulekile kulokhu kuphishekela, ngenxa yekhono labo lokucubungula inani elikhulu ledatha nokubona amaphethini angabonakali abantu abangase bawaphuthelwe ( Using Machine Learning for Stock Market Prediction... | FMP ).

Leli phepha elimhlophe linikeza umbono ophelele wamasu e-AI asetshenziswa ekubikezeleni imakethe yesitoko futhi lihlole ukusebenza kwawo kahle. Sizohlola izisekelo zemfundiso yamamodeli athandwayo (kusukela ezindleleni zendabuko zochungechunge lwesikhathi kuya kumanethiwekhi ajulile e-neural kanye nokufunda kokuqinisa), sixoxe ngenqubo yedatha nokuqeqeshwa kwalawa mamodeli, futhi siqokomise imikhawulo ebalulekile nezinselelo ezibhekene nalezi zinhlelo, njengokusebenza kahle kwemakethe, umsindo wedatha, kanye nemicimbi yangaphandle engalindelekile. Izifundo zangempela nezibonelo zifakiwe ukuze kuboniswe imiphumela exubile etholakale kuze kube manje. Okokugcina, siphetha ngokulindeleka okungokoqobo kubatshalizimali kanye nabasebenzi: siqaphela amakhono ahlaba umxhwele e-AI ngenkathi siqaphela ukuthi izimakethe zezimali zigcina izinga lokungabikezeleki okungekho algorithm elingakuqeda ngokuphelele.

Izisekelo Zemfundiso Ye-AI Ekubikezelweni Kwemakethe Yamasheya

Ukubikezela isitoko kwesimanje okusekelwe ku-AI kwakhelwe phezu kocwaningo lwamashumi eminyaka kwezibalo, ezezimali, kanye nesayensi yamakhompyutha. Kuwusizo ukuqonda ububanzi bezindlela kusukela kumamodeli endabuko kuya ku-AI esezingeni eliphezulu:

• Amamodeli Endabuko Ochungechunge Lwesikhathi: Ukubikezela kwasekuqaleni kwesitoko kuncike kumamodeli ezibalo acabanga ukuthi amaphethini ngamanani adlule angabikezela ikusasa. Amamodeli afana ne- ARIMA (Auto-Regressive Integrated Moving Average) kanye ne -ARCH/GARCH agxile ekubambeni izitayela eziqondile kanye nokuhlanganiswa kokuguquguquka kwedatha yochungechunge lwesikhathi ( Amamodeli okubikezela isitoko aqhutshwa yidatha asekelwe kumanethiwekhi e-neural: Ukubuyekezwa ). Lawa mamodeli ahlinzeka ngesisekelo sokubikezela ngokwenza imodeli yokulandelana kwamanani omlando ngaphansi kokuqagela kokuma kanye nokulingana. Nakuba ewusizo, amamodeli endabuko avame ukuba nenkinga ngamaphethini ayinkimbinkimbi, angewona aqondile ezimakethe zangempela, okuholela ekunembeni okulinganiselwe kokubikezela ekusebenzeni ( Amamodeli okubikezela isitoko aqhutshwa yidatha asekelwe kumanethiwekhi e-neural: Ukubuyekezwa ).

• Ama-Algorithm Okufunda Komshini: Izindlela zokufunda komshini zidlula amafomula ezibalo achazwe ngaphambilini ngokufunda amaphethini ngqo kusuka kudatha . Ama-Algorithm afana nemishini ye-vector yokusekela (i-SVM) , amahlathi angahleliwe , kanye nokukhulisa i-gradient asetshenziswe ekubikezelweni kwesitoko. Angafaka izici eziningi zokufaka - kusukela ezinkomba zobuchwepheshe (isb., izilinganiso ezihambayo, ivolumu yokuhweba) kuya ezinkomba eziyisisekelo (isb., imali engenayo, idatha yezomnotho omkhulu) - futhi athole ubudlelwano obungebona obuqondile phakathi kwawo. Isibonelo, imodeli yokukhulisa ihlathi engahleliwe noma i-gradient ingacabangela izici eziningi ngesikhathi esisodwa, ibambe ukusebenzisana okungaphuthelwa imodeli elula eqondile. Lawa mamodeli e-ML abonise ikhono lokuthuthukisa ngokuthobeka ukunemba kokubikezela ngokuthola izimpawu eziyinkimbinkimbi kudatha ( Ukusebenzisa Ukufunda Komshini Ukubikezela Imakethe Yesitoko... | FMP ). Kodwa-ke, adinga ukulungiswa ngokucophelela kanye nedatha eyanele ukugwema ukufaka ngokweqile (umsindo wokufunda esikhundleni sesignali).

• Ukufunda Okujulile (Amanethiwekhi Emizwa): Amanethiwekhi emizwa ajulile , aphefumulelwe yisakhiwo sobuchopho bomuntu, asethandwa kakhulu ekubikezelweni kwemakethe yamasheya eminyakeni yamuva. Phakathi kwalawa, amaNethiwekhi Emizwa Aphindaphindayo (ama-RNN) kanye namanethiwekhi awo e- Long Short-Term Memory (LSTM) aklanyelwe ngqo idatha yokulandelana njengochungechunge lwesikhathi samanani esitoko. Ama-LSTM angagcina inkumbulo yolwazi oludlule futhi abambe ukuncika kwesikhathi, okwenza afaneleke kahle kumamodeli, imijikelezo, noma amanye amaphethini ancike esikhathini kudatha yemakethe. Ucwaningo lubonisa ukuthi ama-LSTM kanye namanye amamodeli okufunda okujulile angabamba ubudlelwano obuyinkimbinkimbi, obungewona oqondile kudatha yezezimali okungaphuthelwa amamodeli alula. Ezinye izindlela zokufunda okujulile zifaka phakathi amaNethiwekhi Emizwa Aguquguqukayo (ama-CNN) (ngezinye izikhathi asetshenziswa "ezithombeni" zezinkomba zobuchwepheshe noma ukulandelana okufakiwe), ama-Transformers (asebenzisa izindlela zokunaka ukulinganisa ukubaluleka kwezinyathelo zesikhathi ezahlukene noma imithombo yedatha), ngisho namaNethiwekhi Emizwa Angemuva Angemuva (ama-GNN) (ukumodela ubudlelwano phakathi kwamasheya kugrafu yemakethe). Lawa ma-neural network athuthukile angagwinya hhayi nje idatha yamanani kodwa futhi nemithombo yedatha ehlukile njengombhalo wezindaba, imizwa yezokuxhumana, nokuningi, ukufunda izici ezingabonakali ezingase zibikezele ukunyakaza kwemakethe ( Using Machine Learning for Stock Market Prediction... | FMP ). Ukuguquguquka kokufunda okujulile kuza nezindleko: alambele idatha, asebenzisa kakhulu ikhompyutha, futhi avame ukusebenza “njengamabhokisi amnyama” anokuhunyushwa okuncane.

• Ukufunda Kokuqinisa: Omunye umkhawulo ekubikezelweni kwesitoko se-AI ukufunda kokuqinisa (RL) , lapho umgomo ungekhona nje ukubikezela amanani, kodwa nokufunda isu lokuhweba elifanele. Ohlakeni lwe-RL, i -ejenti (imodeli ye-AI) ixhumana nendawo (imakethe) ngokuthatha izinyathelo (ukuthenga, ukuthengisa, ukubamba) nokuthola imivuzo (inzuzo noma ukulahlekelwa). Ngokuhamba kwesikhathi, i-ejenti ifunda inqubomgomo ekhulisa umvuzo oqongelelekayo. Ukufunda Kokuqinisa Okujulile (i-DRL) kuhlanganisa amanethiwekhi e-neural nokufunda kokuqinisa ukuze kusingathwe indawo enkulu yezimakethe. Ukukhanga kwe-RL kwezezimali yikhono layo lokucabangela ukulandelana kwezinqumo nokulungiselela ngqo imbuyiselo yokutshalwa kwezimali, kunokubikezela amanani ngokwawo. Isibonelo, i-ejenti ye-RL ingafunda ukuthi kufanele ingene nini noma iphume ezikhundleni ngokusekelwe kuzimpawu zamanani futhi izivumelanise nezimo njengoba izimo zemakethe zishintsha. Okuphawulekayo ukuthi, i-RL isetshenziswe ukuqeqesha amamodeli e-AI ancintisana emincintiswaneni yokuhweba ngobuningi nakwezinye izinhlelo zokuhweba ezizimele. Kodwa-ke, izindlela ze-RL nazo zibhekene nezinselele ezinkulu: zidinga ukuqeqeshwa okubanzi (ukulingisa iminyaka yokuhweba), zingahlushwa ukungazinzi noma ukuziphatha okuhlukile uma zingalungiswanga kahle, futhi ukusebenza kwazo kuzwela kakhulu endaweni yemakethe ecatshangelwayo. Abacwaningi baphawule izinkinga ezifana nezindleko eziphezulu zokubala nezinkinga zokuzinza ekusebenziseni ukufunda kokuqinisa ezimakethe zamasheya eziyinkimbinkimbi. Naphezu kwalezi zinselele, i-RL imelela indlela ethembisayo, ikakhulukazi uma ihlanganiswa namanye amasu (isb., ukusebenzisa amamodeli okubikezela intengo kanye nesu lokwabiwa okusekelwe ku-RL) ukwakha uhlelo lokwenza izinqumo oluhlanganisiwe ( Ukubikezela Imakethe Yamasheya Ukusebenzisa Ukufunda Okujulile Kokuqinisa ).

Imithombo Yedatha kanye Nenqubo Yokuqeqesha

Kungakhathaliseki uhlobo lwemodeli, idatha iyinsika yokubikezela imakethe yesitoko se-AI. Amamodeli avame ukuqeqeshwa ngedatha yemakethe yomlando kanye namanye amasethi edatha ahlobene ukuthola amaphethini. Imithombo yedatha evamile nezici zifaka:

• Amanani Omlando Nezinkomba Zobuchwepheshe: Cishe wonke amamodeli asebenzisa amanani esitoko adlule (avulekile, aphezulu, aphansi, aseduze) kanye namanani okuhweba. Kulawa, abahlaziyi bavame ukuthola izinkomba zobuchwepheshe (izilinganiso ezihambayo, inkomba yamandla ahlobene, i-MACD, njll.) njengezinto ezifakiwe. Lezi zinkomba zingasiza ekugqamiseni izitayela noma umfutho imodeli engase iwusebenzise. Isibonelo, imodeli ingase ithathe njengokufaka ezinsukwini zokugcina eziyi-10 zamanani kanye nomthamo, kanye nezinkomba ezifana nezilinganiso zokuhamba zezinsuku eziyi-10 noma izindlela zokuguquguquka, ukubikezela ukunyakaza kwentengo kosuku olulandelayo.

• Izinkomba Zemakethe Nedatha Yezomnotho: Amamodeli amaningi afaka ulwazi olubanzi lwemakethe, njengamazinga ezinkomba, amazinga enzalo, ukukhuphuka kwamanani entengo, ukukhula kwe-GDP, noma ezinye izinkomba zezomnotho. Lezi zici ezinkulu zinikeza umongo (isb., umuzwa wemakethe noma impilo yezomnotho) ongathonya ukusebenza kwesitoko somuntu ngamunye.

• Idatha Yezindaba Nemizwa: Inani elikhulayo lezinhlelo ze-AI lidla idatha engahlelekile njengezihloko zezindaba, okuphakelayo kwezokuxhumana (i-Twitter, i-Stocktwits), kanye nemibiko yezezimali. Amasu Okucubungula Ulimi Lwemvelo (i-NLP), kufaka phakathi amamodeli athuthukile njenge-BERT, asetshenziselwa ukulinganisa imizwa yemakethe noma ukuthola imicimbi efanele. Isibonelo, uma imizwa yezindaba ngokuzumayo iba yimbi kakhulu enkampanini noma emkhakheni, imodeli ye-AI ingase ibikezele ukwehla kwamanani esitoko ahlobene. Ngokucubungula izindaba zesikhathi sangempela kanye nemizwa yezokuxhumana , i-AI ingasabela ngokushesha kunabathengisi abangabantu olwazini olusha.

• Idatha Ehlukile: Amanye abacwaningi be-hedge funds abanolwazi olunzulu kanye ne-AI basebenzisa imithombo yedatha ehlukile - izithombe zesathelayithi (zethrafikhi yesitolo noma imisebenzi yezimboni), idatha yokuthengiselana ngamakhadi esikweletu, izitayela zokusesha iwebhu, njll. - ukuthola ukuqonda okubikezelayo. Lawa masethi edatha angewona avamile ngezinye izikhathi angasebenza njengezinkomba ezihamba phambili zokusebenza kwesitoko, yize futhi ethula ubunzima ekuqeqeshweni kwamamodeli.

Ukuqeqesha imodeli ye-AI yokubikezela isitoko kuhilela ukuyipha le datha yomlando nokulungisa amapharamitha emodeli ukuze kuncishiswe iphutha lokubikezela. Ngokuvamile, idatha ihlukaniswe yaba isethi yokuqeqesha (isb., umlando omdala wokufunda amaphethini) kanye nesethi yokuhlola/yokuqinisekisa (idatha yakamuva yokuhlola ukusebenza ezimweni ezingabonakali). Njengoba kunikezwe uhlobo lokulandelana kwedatha yemakethe, kuyacatshangelwa ukugwema "ukubheka esikhathini esizayo" - isibonelo, amamodeli ahlolwa kudatha kusukela ezikhathini zesikhathi ngemva kwesikhathi sokuqeqeshwa, ukuze alingise ukuthi azosebenza kanjani ekuhwebeni kwangempela. okuqinisekisa aqondiswe ochungechungeni lwesikhathi (njengokuqinisekiswa kokuhamba phambili) asetshenziswa ukuqinisekisa ukuthi imodeli ihlangana kahle futhi ayifaneleki nje kuphela esikhathini esithile esithile.

Ngaphezu kwalokho, ochwepheshe kumele babhekane nezinkinga zekhwalithi yedatha kanye nokucubungula kwangaphambili. Idatha engekho, izinto ezingaphandle (isb., ukwanda okungazelelwe ngenxa yokuhlukaniswa kwesitoko noma izenzakalo zesikhathi esisodwa), kanye nezinguquko zombuso ezimakethe konke kungathinta ukuqeqeshwa kwemodeli. Amasu anjengokwenza kube ngokwejwayelekile, ukuqeda izitayela, noma ukuqeda izinkathi angasetshenziswa kudatha yokufaka. Ezinye izindlela ezithuthukisiwe zihlukanisa uchungechunge lwamanani zibe yizingxenye (izitayela, imijikelezo, umsindo) bese ziwaklama ngokwehlukana (njengoba kubonwe ocwaningweni oluhlanganisa ukubola kwemodi eguquguqukayo namanethi e-neural ( Ukubikezela Imakethe Yesitoko Ukusebenzisa Ukufunda Okujulile Kokuqinisa )).

Amamodeli ahlukene anezidingo zokuqeqeshwa ezihlukene: amamodeli okufunda okujulile angadinga amakhulu ezinkulungwane zamaphuzu edatha futhi azuze ekusheshisweni kwe-GPU, kanti amamodeli alula njengokuhlehla kwe-logistic angafunda kumasethi edatha amancane kakhulu. Amamodeli okufunda okuqinisa adinga isilingisi noma indawo yokuxhumana nayo; ngezinye izikhathi idatha yomlando iphindwa ku-ejenti ye-RL, noma izilingisi zemakethe zisetshenziselwa ukukhiqiza okuhlangenwe nakho.

Okokugcina, uma seziqeqeliwe, lezi zinhlobo ziveza umsebenzi wokubikezela - isibonelo, umkhiqizo ongaba yintengo ebikezelwe yakusasa, amathuba okuthi isitoko sizokhuphuka, noma isenzo esinconywayo (ukuthenga/ukuthengisa). Lezi zibikezelo zivame ukuhlanganiswa nesu lokuhweba (ngosayizi wesikhundla, imithetho yokuphatha ubungozi, njll.) ngaphambi kokuba imali yangempela ibekwe engcupheni.

Ukulinganiselwa kanye Nezinselele

Nakuba amamodeli e-AI esethuthuke kakhulu, ukubikezela imakethe yamasheya kusalokhu kungumsebenzi oyinselele ngokwemvelo . Lokhu okulandelayo yimikhawulo ebalulekile kanye nezithiyo ezivimbela i-AI ekubeni ngumbikezeli wenhlanhla oqinisekisiwe emakethe:

• Ukusebenza Kahle Kwemakethe Nokungahleliwe: Njengoba kushiwo ngaphambili, i-Efficient Market Hypothesis ithi amanani asevele ebonakalisa ulwazi olwaziwayo, ngakho-ke noma yiluphi ulwazi olusha lubangela ukulungiswa okusheshayo. Ngamagama asebenzayo, lokhu kusho ukuthi izinguquko zentengo ziqhutshwa kakhulu yizindaba ezingalindelekile noma ukuguquguquka okungahleliwe. Ngempela, amashumi eminyaka ocwaningo athole ukuthi ukunyakaza kwentengo yesitoko yesikhathi esifushane kufana nokuhamba okungahleliwe ( Amamodeli okubikezela isitoko aqhutshwa idatha asekelwe kumanethiwekhi e-neural: Ukubuyekezwa ) - intengo yangaphambilini ayinamthelela omkhulu kweyakusasa, ngale kwalokho okungenzeka ukuthi kungabikezelwa. Uma amanani esitoko empeleni engahleliwe noma "esebenza kahle," ayikho i-algorithm engayibikezela njalo ngokunemba okuphezulu. Njengoba olunye ucwaningo lukubeka kafushane, "i-random walk hypothesis kanye ne-effective market hypothesis empeleni ithi akunakwenzeka ukubikezela ngokuhlelekile nangokuthembekile amanani esitoko esizayo" ( Ukubikezela imbuyiselo ehlobene yamasheya e-S&P 500 kusetshenziswa ukufunda komshini | Ukuqamba Kwezezimali | Umbhalo Ogcwele ). Lokhu akusho ukuthi izibikezelo ze-AI zihlala zingasizi ngalutho, kodwa kugcizelela umkhawulo oyisisekelo: iningi lokunyakaza kwemakethe kungaba umsindo nje ngisho nemodeli engcono kakhulu engenakubikezela kusengaphambili.

• Umsindo Nezici Zangaphandle Ezingabikezeleki: Amanani esitoko athonywa yizici eziningi, eziningi zazo ezingaphandle futhi ezingabikezeleki. Imicimbi yezepolitiki (izimpi, ukhetho, izinguquko zomthetho), izinhlekelele zemvelo, ubhubhane, amahlazo ezinkampani angazelelwe, noma ngisho namahemuhemu ezinkundla zokuxhumana asakazekile konke kungashukumisa izimakethe ngokungazelelwe. Lezi yimicimbi lapho imodeli ingenakuba nedatha yokuqeqeshwa kwangaphambilini (ngoba ayikaze ibonwe ngaphambili) noma eyenzeka njengokushaqeka okungavamile. Isibonelo, ayikho imodeli ye-AI eqeqeshwe ngedatha yomlando kusukela ngo-2010-2019 eyayingabikezela ngokuqondile ukuwa kwe-COVID-19 ekuqaleni kuka-2020 noma ukubuya kwayo ngokushesha. Amamodeli e-AI ezezimali ayalwa lapho ohulumeni beshintsha noma lapho isenzakalo esisodwa siqhuba amanani. Njengoba omunye umthombo ephawula, izici ezifana nemicimbi yezepolitiki noma ukukhishwa kwedatha yezomnotho okungazelelwe kungenza izibikezelo ziphelelwe yisikhathi cishe ngokushesha ( Ukusebenzisa i-Machine Learning for Stock Market Prediction... | FMP ) ( Ukusebenzisa i-Machine Learning for Stock Market Prediction... | FMP ). Ngamanye amazwi, izindaba ezingalindelekile zingahlala zidlula izibikezelo ze-algorithmic , zifake izinga lokungaqiniseki elingenakwehliswa.

• Ukufakwa ngokweqile kanye nokuhlanganiswa: Amamodeli okufunda komshini athambekele ekufakweni ngokweqile - okusho ukuthi angafunda "umsindo" noma izimo ezingavamile kudatha yokuqeqeshwa kahle kakhulu, kunokuba afunde amaphethini ajwayelekile ayisisekelo. Imodeli efakwe ngokweqile ingase isebenze kahle kakhulu kudatha yomlando (ngisho nokubonisa imbuyiselo emangalisayo evivinywe emuva noma ukunemba okuphezulu kwesampula) kodwa bese yehluleka kabi kudatha entsha. Lokhu kuyisithiyo esivamile kwezezimali ezilinganiselwe. Isibonelo, inethiwekhi ye-neural eyinkimbinkimbi ingase ithole ubudlelwano obungamanga obabambe isikhathi esidlule ngengozi (njengenhlanganisela ethile yezinkomba ezenzeka ngaphambi kwemihlangano eminyakeni emi-5 edlule) kodwa lobo budlelwano bungase bungabi khona kusukela phambili. Umfanekiso osebenzayo: umuntu angaklama imodeli ebikezela ukuthi abawinile besitoko sonyaka odlule bazohlala bekhuphuka - ingase ivumelane nesikhathi esithile, kodwa uma uhlelo lwemakethe lushintsha, lelo phethini liyaphuka. Ukufakwa ngokweqile kuholela ekusebenzeni okubi ngaphandle kwesampula , okusho ukuthi izibikezelo zemodeli ekuhwebeni bukhoma azinakuba ngcono kunokungahleliwe naphezu kokubukeka okuhle ekuthuthukisweni. Ukugwema ukufakwa ngokweqile kudinga amasu afana nokuhlelwa kabusha, ukugcina ubunzima bemodeli bulawulwa, kanye nokusebenzisa ukuqinisekiswa okuqinile. Kodwa-ke, ubunzima obunikeza amamodeli e-AI amandla bubenza babe sengozini yalolu daba.

• Ikhwalithi Yedatha Nokutholakala Kwayo: Isisho esithi “udoti ungene, udoti uphume” sisebenza kakhulu ekubikezelweni kwesitoko. Ikhwalithi, ubuningi, kanye nokufaneleka kwedatha kuthinta kakhulu ukusebenza kwemodeli. Uma idatha yomlando inganele (isb., ukuzama ukuqeqesha inethiwekhi ejulile ngeminyaka embalwa nje yamanani esitoko) noma ukungameleli (isb., ukusebenzisa idatha yesikhathi esibi kakhulu ukubikezela isimo esibi), imodeli ngeke ihlanganise kahle. Idatha ingabuye ibe nobandlululo noma incike ekusindeni (isibonelo, izinkomba zesitoko ngokwemvelo ziwisa izinkampani ezingasebenzi kahle ngokuhamba kwesikhathi, ngakho idatha yenkomba yomlando ingase ibe nobandlululo phezulu). Ukuhlanza nokugcina idatha kuwumsebenzi ongelula. Ngaphezu kwalokho, eminye imithombo yedatha ingaba ebiza kakhulu noma kube nzima ukuyithola, okungase kunikeze abadlali bezikhungo ithuba ngenkathi kushiya abatshalizimali bezitolo benedatha engaphelele. Kukhona futhi nenkinga yemvamisa : amamodeli okuhweba avame kakhulu adinga idatha ye-tick-by-tick enkulu kakhulu futhi edinga ingqalasizinda ekhethekile, kanti amamodeli avame kakhulu angasebenzisa idatha yansuku zonke noma yamasonto onke. Ukuqinisekisa ukuthi idatha ihambisana nesikhathi (isb., izindaba nedatha yamanani ahambisanayo) futhi ayinakho ukucwasa okubonakalayo kuyinselele eqhubekayo.

• Ukucaca Nokuhunyushwa Kwemodeli: Amamodeli amaningi e-AI, ikakhulukazi lawo afunda ngokujulile, asebenza njengamabhokisi amnyama . Angase aveze isibikezelo noma isignali yokuhweba ngaphandle kwesizathu esilula ukusichaza. Lokhu kuntuleka kokucaca kungaba yinkinga kubatshalizimali - ikakhulukazi labo abasungula izikhungo abadinga ukuthethelela izinqumo kubabambiqhaza noma ukulandela imithetho. Uma imodeli ye-AI ibikezela ukuthi isitoko sizowa futhi incoma ukuthengiswa, umphathi wephothifoliyo angase angabaze uma engaqondi isizathu. Ukungacaci kwezinqumo ze-AI kunganciphisa ukwethenjwa nokwamukelwa, kungakhathaliseki ukuthi imodeli inembile kangakanani. Le nselele ikhuthaza ucwaningo nge-AI echazekayo yezezimali, kodwa kuyiqiniso ukuthi kuvame ukuhwebelana phakathi kobunzima/ukunemba kwemodeli kanye nokuhunyushwa kwayo.

• Izimakethe Ezivumelana Nezimo Nokuncintisana: Kubalulekile ukuqaphela ukuthi izimakethe zezimali ziyavumelana nezimo . Uma iphethini yokubikezela isitholiwe (nge-AI noma nganoma iyiphi indlela) futhi isetshenziswa abadayisi abaningi, ingase iyeke ukusebenza. Isibonelo, uma imodeli ye-AI ithola ukuthi isignali ethile ivame ngaphambi kokukhuphuka kwesitoko, abadayisi bazoqala ukwenza leso signali kusenesikhathi, ngaleyo ndlela baxazulule ithuba. Empeleni, izimakethe zingathuthuka ukuze ziqede amasu aziwayo . Namuhla, izinkampani eziningi zokuhweba kanye nezimali zisebenzisa i-AI ne-ML. Lo mncintiswano usho ukuthi noma yimuphi umkhawulo uvame ukuba mncane futhi uhlala isikhathi esifushane. Umphumela uba ukuthi amamodeli e-AI angadinga ukuqeqeshwa kabusha nokubuyekezwa njalo ukuze ahambisane nokushintshashintsha kwemakethe. Ezimakethe ezimanzi kakhulu nezivuthiwe (njengamasheya amakhulu ase-US), abadlali abaningi abanolwazi bafuna izimpawu ezifanayo, okwenza kube nzima kakhulu ukugcina umkhawulo. Ngokuphambene nalokho, ezimakethe ezingasebenzi kahle noma ezimpahleni ze-niche, i-AI ingase ithole ukungasebenzi kahle kwesikhashana - kodwa njengoba lezo zimakethe zishintsha, igebe lingavaleka. Lolu hlobo lokuguquguquka kwezimakethe luyinselele eyisisekelo: "imithetho yomdlalo" ayimi ndawonye, ​​​​ngakho imodeli esebenza ngonyaka odlule ingadinga ukulungiswa kabusha ngonyaka ozayo.

• Izithiyo Zomhlaba Wangempela: Ngisho noma imodeli ye-AI ingabikezela amanani ngokunemba okuhle, ukuguqula izibikezelo zibe yinzuzo kungenye inselele. Ukuhweba kubangela izindleko zokuthengiselana , njengekhomishini, ukwehla, kanye nezintela. Imodeli ingabikezela ukunyakaza okuningi kwamanani amancane ngendlela efanele, kodwa izinzuzo zingasuswa yizimali kanye nomthelela wemakethe wokuhweba. Ukuphathwa kwengozi nakho kubalulekile - akukho ukubikezela okuqinisekile ngo-100%, ngakho-ke noma yiliphi isu eliqhutshwa yi-AI kumele libhekele ukulahlekelwa okungenzeka (ngokusebenzisa ama-oda okulahlekelwa yi-stop, ukuhlukahluka kwephothifoliyo, njll.). Izikhungo zivame ukuhlanganisa izibikezelo ze-AI ohlakeni olubanzi lwengozi ukuqinisekisa ukuthi i-AI ayibheki ipulazi ngesibikezelo esingaba yiphutha. Lokhu kucatshangelwa okusebenzayo kusho ukuthi umkhawulo wethiyori we-AI kumele ube mkhulu ukuze ube wusizo ngemuva kokungqubuzana komhlaba wangempela.

Ngamafuphi, i-AI inamakhono amakhulu, kodwa le mikhawulo iqinisekisa ukuthi imakethe yamasheya ihlala iyisistimu ebikezelwayo kancane, engabikezelwa kancane . Amamodeli e-AI anganciphisa amathuba ukuze athole umtshali-zimali ngokuhlaziya idatha ngempumelelo futhi mhlawumbe embule izimpawu zokubikezela ezicashile. Kodwa-ke, inhlanganisela yamanani asebenzayo, idatha enomsindo, izehlakalo ezingalindelekile, kanye nemikhawulo engokoqobo kusho ukuthi ngisho ne-AI engcono kakhulu ngezinye izikhathi izoba nephutha - ngokuvamile ngendlela engabikezelwanga.

Ukusebenza Kwamamodeli E-AI: Buthini Ubufakazi?

Uma sibheka kokubili intuthuko kanye nezinselele okuxoxwe ngazo, yini esiyifundile ocwaningweni kanye nemizamo yangempela yokusebenzisa i-AI ekubikezelweni kwesitoko? Imiphumela kuze kube manje ixubile, igqamisa impumelelo ethembisayo kanye nokwehluleka okuxakayo :

• Izimo Zethuba Elihle Kakhulu Le-AI: Izifundo eziningana zibonise ukuthi amamodeli e-AI anganqoba ukuqagela okungahleliwe ngaphansi kwezimo ezithile. Isibonelo, ucwaningo lwango-2024 lusebenzise inethiwekhi ye-LSTM neural ukubikezela izitayela emakethe yesitoko yaseVietnam futhi lwabika ukunemba okuphezulu kokubikezela - cishe ama-93% kudatha yokuhlola ( Ukusebenzisa ama-algorithms okufunda komshini ukubikezela izitayela zentengo yesitoko emakethe yesitoko - Icala laseVietnam | Ukuxhumana Kwezenhlalo Nesayensi Yezenhlalo ). Lokhu kusikisela ukuthi kuleyo makethe (umnotho osafufusa), imodeli yakwazi ukubamba amaphethini ahambisanayo, mhlawumbe ngoba imakethe yayinezinkinga noma izitayela zobuchwepheshe eziqinile ezifundwe yi-LSTM. Olunye ucwaningo ngo-2024 lwathatha ububanzi obanzi: abacwaningi bazama ukubikezela imbuyiselo yesikhashana yazo zonke izitoko ze-S&P 500 (imakethe esebenza kahle kakhulu) besebenzisa amamodeli e-ML. Bakubeke njengenkinga yokuhlukaniswa - ukubikezela ukuthi isitoko sizodlula yini inkomba ngo-2% ezinsukwini eziyi-10 ezizayo - besebenzisa ama-algorithms afana ne-Random Forests, SVM, kanye ne-LSTM. Umphumela: imodeli ye-LSTM iphumelele kakhulu kunezinye izinhlobo ze-ML kanye nesisekelo esingahleliwe , nemiphumela ibaluleke ngokwanele ngokwezibalo ukusikisela ukuthi kwakungeyona nje inhlanhla ( Ukubikezela imbuyiselo ehlobene yamasheya e-S&P 500 kusetshenziswa ukufunda komshini | Ukuqamba Kwezezimali | Umbhalo Ogcwele ). Ababhali baze baphetha ngokuthi kulokhu kusethwa okuthile, amathuba okuthi i- random walk hypothesis iwaphethe "ayemancane kakhulu," okubonisa ukuthi amamodeli abo e-ML athole izimpawu zangempela zokubikezela. Lezi zibonelo zibonisa ukuthi i-AI ingabona ngempela amaphethini anikeza umkhawulo (noma ngabe uphansi) ekubikezeleni ukuhamba kwesitoko, ikakhulukazi uma kuhlolwa kumasethi amakhulu edatha.

• Izimo Zokusetshenziswa Eziphawulekayo Embonini: Ngaphandle kwezifundo zemfundo, kunemibiko yezimali ezithwala imithwalo yemfanelo kanye nezikhungo zezimali ezisebenzisa i-AI ngempumelelo emisebenzini yazo yokuhweba. Ezinye izinkampani zokuhweba ezivame kakhulu zisebenzisa i-AI ukuqaphela nokusabela kumaphethini esakhiwo esincane semakethe ngamaqhezu omzuzwana. Amabhange amakhulu anamamodeli e-AI okwabiwa kwephothifoliyo kanye nokubikezela ubungozi , okuthi, yize kungahlali kumayelana nokubikezela intengo yesitoko esisodwa, kuhilela ukubikezela izici zemakethe (njengokuguquguquka noma ukuhlangana). Kukhona futhi izimali eziqhutshwa yi-AI (ezivame ukubizwa ngokuthi “izimali ze-quant”) ezisebenzisa ukufunda komshini ukwenza izinqumo zokuhweba - ezinye zisebenze kakhulu kunemakethe isikhathi esithile, yize kunzima ukusho lokho ku-AI ngoba zivame ukusebenzisa inhlanganisela yobuhlakani bomuntu nobomshini. Isicelo esiqondile ukusetshenziswa kwe yokuhlaziya imizwa : isibonelo, ukuskena izindaba kanye ne-Twitter ukubikezela ukuthi amanani esitoko azoshintsha kanjani ngokuphendula. Amamodeli anjalo angase angabi neqiniso elingu-100%, kodwa anganikeza abahwebi isiqalo esincane samanani ezindabeni. Kubalulekile ukuqaphela ukuthi izinkampani ngokuvamile ziqapha imininingwane yamasu e-AI aphumelelayo njengempahla yobuhlakani, ngakho-ke ubufakazi obusemphakathini buvame ukubambezeleka noma bube yindaba engasho lutho.

• Amacala Okungasebenzi Kahle Nokwehluleka: Kuzo zonke izindaba zempumelelo, kunezindaba zokuxwayisa. Izifundo eziningi zezemfundo ezithi zinembile kakhulu emakethe eyodwa noma ngesikhathi esisodwa zehlulekile ukuhlanganisa. Ukuhlolwa okuphawulekayo kwazama ukuphinda isifundo esiphumelelayo sokubikezela imakethe yesitoko saseNdiya (esasinokunemba okuphezulu sisebenzisa i-ML ezinkombeni zobuchwepheshe) ezitokisini zase-US. Ukuphindaphinda akutholanga amandla abalulekile okubikezela - empeleni, isu elingenalwazi lokubikezela njalo ukuthi isitoko sizokhuphuka ngosuku olulandelayo ladlula amamodeli e-ML ayinkimbinkimbi ngokunemba. Abalobi baphetha ngokuthi imiphumela yabo "isekela inkolelo-mbono yokuhamba ngokungahleliwe" , okusho ukuthi ukunyakaza kwesitoko bekungabikezeleki futhi amamodeli e-ML awazange asize. Lokhu kugcizelela ukuthi imiphumela ingahluka kakhulu ngokwemakethe nangesikhathi. Ngokufanayo, imincintiswano eminingi ye-Kaggle kanye nemincintiswano yocwaningo lwe-quant ikhombisile ukuthi yize amamodeli evame ukulingana kahle nedatha edlule, ukusebenza kwawo ekuhwebeni bukhoma kuvame ukuhlehlela emuva ekunembileni okungu-50% (ukubikezela isiqondiso) uma ebhekene nezimo ezintsha. Izimo ezifana nokuwa kwesikhwama se-quant sika-2007 kanye nobunzima obubhekene nezimali eziqhutshwa yi-AI ngesikhathi sokushaqeka kobhubhane luka-2020 zibonisa ukuthi amamodeli e-AI angawa ngokuzumayo lapho umbuso wemakethe ushintsha. Ubandlululo lokusinda luyisici esibalulekile emibonweni - sizwa ngempumelelo ye-AI kaningi kunokwehluleka, kodwa ngemuva kwezigcawu, amamodeli amaningi kanye nezimali ziyahluleka futhi zivalwe buthule ngoba amasu azo ayayeka ukusebenza.

• Umehluko Kuwo Wonke Amamakethe: Okuphawuliwe okuthakazelisayo okuvela ezifundweni ukuthi ukusebenza kahle kwe-AI kungancika ekuvuthweni kwemakethe kanye nokusebenza kahle . Ezimakethe ezingasebenzi kahle noma ezisafufusa, kungase kube namaphethini angasetshenziswa kakhulu (ngenxa yokumbozwa okuphansi kwabahlaziyi, imikhawulo yoketshezi, noma ukucwasa kokuziphatha), okuvumela amamodeli e-AI ukuthi afinyelele ukunemba okuphezulu. Ucwaningo lwemakethe yaseVietnam lwe-LSTM olune-93% lungaba yisibonelo salokhu. Ngokuphambene nalokho, ezimakethe ezisebenza kahle kakhulu njenge-US, lawo maphethini angase axazululwe ngokushesha. Imiphumela exubile phakathi kwecala laseVietnam kanye nocwaningo lokuphindaphinda lwase-US lubonisa lokhu kungafani. Emhlabeni jikelele, lokhu kusho ukuthi i-AI okwamanje ingase iveze ukusebenza okungcono kokubikezela ezimakethe ezithile ze-niche noma ezigabeni zezimpahla (isibonelo, abanye basebenzise i-AI ukubikezela amanani ezimpahla noma izitayela ze-cryptocurrency ngempumelelo ehlukahlukene). Ngokuhamba kwesikhathi, njengoba zonke izimakethe ziqhubekela ekusebenzeni kahle okukhulu, ifasitela lokuwina okulula kokubikezela liyancipha.

• Ukunemba vs. Inzuzo: Kubalulekile futhi ukuhlukanisa ukunemba kokubikezela nenzuzo yokutshalwa kwezimali . Imodeli ingaba neqiniso elingu-60% kuphela ekubikezeleni ukunyakaza kwansuku zonke kokukhuphuka noma ukwehla kwesitoko - okungezwakali kuphezulu kakhulu - kodwa uma lezo zibikezelo zisetshenziswa isu lokuhweba elihlakaniphile, zingaba nenzuzo enkulu. Ngakolunye uhlangothi, imodeli ingase ibe nokunemba okungu-90% kodwa uma izikhathi ezingu-10% ingalungile zihambisana nokunyakaza okukhulu kwemakethe (futhi ngaleyo ndlela ukulahlekelwa okukhulu), kungaba yinto engenanzuzo. Imizamo eminingi yokubikezela isitoko se-AI igxila ekuqondeni okuqondile noma ekunciphiseni amaphutha, kodwa abatshalizimali bayakhathalela ukubuyiselwa okulungisiwe kwengozi. Ngakho-ke, ukuhlolwa kuvame ukufaka izilinganiso ezifana nesilinganiso se-Sharpe, ukwehla, kanye nokuvumelana kokusebenza, hhayi nje izinga lokushaya elingavuthiwe. Amanye amamodeli e-AI ahlanganiswe ezinhlelweni zokuhweba ze-algorithmic eziphatha izikhundla kanye nengozi ngokuzenzakalelayo - ukusebenza kwawo kwangempela kulinganiswa ekubuyiselweni kokuhweba okubukhoma kunezibalo zokubikezela ezizimele. Kuze kube manje, "umhwebi we-AI" ozimele ngokuphelele okhiqiza imali ngokuthembekile unyaka nonyaka uyinganekwane yesayensi kakhulu kuneqiniso, kodwa izinhlelo zokusebenza ezincane (njengemodeli ye-AI ebikezela ukuguquguquka abahwebi abangayisebenzisa ukuze bathenge izinketho, njll.) bathole indawo kuthuluzi lezezimali.

Sekukonke, ubufakazi busikisela ukuthi i-AI ingabikezela amaphethini athile emakethe ngokunemba okungcono kunokungalindelekile , futhi ngokwenza kanjalo kunganikeza inzuzo yokuhweba. Kodwa-ke, leyo nzuzo ivame ukuba ncane futhi idinga ukwenziwa okuyinkimbinkimbi ukuze kuzuzwe inzuzo. Uma umuntu ebuza, ingabe i-AI ingabikezela imakethe yamasheya?, impendulo eqotho kakhulu esekelwe ebufakazini bamanje ithi: I-AI ngezinye izikhathi ingabikezela izici zemakethe yamasheya ngaphansi kwezimo ezithile, kodwa ayikwazi ukwenza kanjalo ngokuqhubekayo kuzo zonke izitoko ngaso sonke isikhathi . Impumelelo ivame ukuba yingxenye futhi incike kumongo.

Isiphetho: Amathemba Angokoqobo E-AI Ekubikezelweni Kwemakethe Yamasheya

I-AI kanye nokufunda komshini ngokungangabazeki kube amathuluzi anamandla kwezezimali. Bahamba phambili ekucubunguleni amasethi edatha amakhulu, bembula ukuxhumana okufihliwe, ngisho nokuguqula amasu ngokushesha. Emizamweni yokubikezela imakethe yamasheya, i-AI ilethe okubonakalayo kodwa okulinganiselwe . Abatshalizimali kanye nezikhungo bangalindela ngempela ukuthi i-AI isize ekwenzeni izinqumo - isibonelo, ngokukhiqiza izimpawu zokubikezela, ukwenza ngcono amaphothifoliyo, noma ukuphatha ubungozi - kodwa hhayi ukusebenza njengebhola lekristalu eliqinisekisa inzuzo.

Lokho i-AI
engakwenza : I-AI ingathuthukisa inqubo yokuhlaziya ekutshalweni kwezimali. Ingahlunga iminyaka yedatha yemakethe, okuphakelayo kwezindaba, kanye nemibiko yezezimali ngemizuzwana, ithole amaphethini angabonakali noma izinto ezingavamile umuntu angase azikhohlwe ( Using Machine Learning for Stock Market Prediction... | FMP ). Ingahlanganisa amakhulu eziguquguquko (ezobuchwepheshe, eziyisisekelo, ezithinta imizwa, njll.) ibe yisibikezelo esihlangene. Ekuhwebeni kwesikhathi esifushane, ama-algorithm e-AI angabikezela ngokunemba okungcono kunokungahleliwe ukuthi isitoko esisodwa sizodlula esinye, noma ukuthi imakethe isizobhekana nokwanda kokuguquguquka. Lezi ziphetho ezikhuphukayo, uma zisetshenziswa kahle, zingahumushela ezinzuzweni zangempela zezimali. I-AI ingasiza futhi ekuphathweni kwezingozi - ukuhlonza izixwayiso zakuqala zokwehla noma ukwazisa abatshalizimali ngezinga lokuzethemba lesibikezelo. Enye indima esebenzayo ye-AI iwukwenza okuzenzakalelayo kwesu : ama-algorithm angenza ukuhweba ngesivinini esikhulu kanye nokuphindaphinda, aphendule ezenzakalweni amahora angu-24 ngosuku, izinsuku eziyi-7 ngesonto, futhi aphoqelele isiyalo (akukho ukuhweba ngokomzwelo), okungaba nenzuzo ezimakethe ezishintshashintshayo.

Lokho i-AI
engenakukwenza (Okwamanje): Naphezu kokuduma kweminye imithombo yezindaba, i-AI ayikwazi ukubikezela njalo nangokuthembekile imakethe yamasheya ngomqondo ophelele wokuhlala inqoba imakethe noma ukubona kusengaphambili amaphuzu amakhulu okuguquka. Izimakethe zithintwa ukuziphatha kwabantu, izehlakalo ezingahleliwe, kanye nezimo eziyinkimbinkimbi zempendulo eziphikisana nanoma iyiphi imodeli engaguquki. I-AI ayisusi ukungaqiniseki; isebenza kuphela ngamathuba. I-AI ingabonisa amathuba angu-70% okuthi isitoko sizokhuphuka kusasa - okusho futhi amathuba angu-30% okungeke kwenzeke. Ukulahlekelwa ukuhweba kanye nezingcingo ezimbi akunakugwenywa. I-AI ayikwazi ukulindela imicimbi emisha ngempela (evame ukubizwa ngokuthi “ama-swans amnyama”) angaphandle kwedatha yayo yokuqeqeshwa. Ngaphezu kwalokho, noma iyiphi imodeli yokubikezela ephumelelayo imema umncintiswano onganciphisa inzuzo yawo. Empeleni, ayikho i-AI efana nebhola lekristalu eliqinisekisa ukubona kusengaphambili ikusasa lemakethe. Abatshalizimali kufanele baqaphele noma ubani othi ngenye indlela.

Umbono Ongathathi hlangothi Nongokoqobo:
Ngokombono ongathathi hlangothi, i-AI ibhekwa kangcono njengokuthuthukiswa, hhayi njengokufaka esikhundleni, ukuhlaziywa kwendabuko kanye nokuqonda komuntu. Empeleni, abatshalizimali abaningi bezikhungo basebenzisa amamodeli e-AI kanye nokufakwa okuvela kubahlaziyi babantu kanye nabaphathi bephothifoliyo. I-AI ingase inciphise izinombolo kanye nezibikezelo zokukhipha, kodwa abantu babeka izinhloso, bahumushe imiphumela, futhi balungise amasu omongo (isb., ukweqa imodeli ngesikhathi senhlekelele engalindelekile). Abatshalizimali bezitolo abasebenzisa amathuluzi aqhutshwa yi-AI noma ama-bot okuhweba kufanele bahlale beqaphile futhi baqonde ukucabanga kwethuluzi kanye nemikhawulo yalo. Ukulandela isincomo se-AI ngokungacabangi kuyingozi - umuntu kufanele ayisebenzise njengokufakwa okukodwa phakathi kwabaningi.

Ekubekeni amathemba angokoqobo, umuntu angaphetha ngokuthi: I-AI ingabikezela imakethe yamasheya ngezinga elithile, kodwa hhayi ngokuqiniseka futhi hhayi ngaphandle kwephutha . Ingandisa amathuba okwenza ucingo olufanele noma ithuthukise ukusebenza kahle ekuhlaziyeni ulwazi, okungenza umehluko phakathi kwenzuzo nokulahlekelwa ezimakethe ezincintisanayo. Kodwa-ke, ayikwazi ukuqinisekisa impumelelo noma ukuqeda ukuguquguquka okungokwemvelo kanye nengozi yezimakethe zamasheya. Njengoba enye incwadi ibonisile, ngisho nangama-algorithms asebenzayo, imiphumela emakethe yamasheya ingaba "engalindelekile ngokwemvelo" ngenxa yezici ezingaphezu kolwazi oluhleliwe ( Ukubikezela Imakethe Yamasheya Ukusebenzisa Ukufunda Okujulile Kokuqinisa ).

Indlela Eya Phambili:
Uma sibheka phambili, indima ye-AI ekubikezelweni kwemakethe yesitoko cishe izokhula. Ucwaningo oluqhubekayo lubhekana neminye yemikhawulo (isibonelo, ukuthuthukisa amamodeli abheka izinguquko zombuso, noma izinhlelo ezihlanganisiwe ezifaka kokubili ukuhlaziywa okuqhutshwa idatha kanye nokuqhutshwa yimicimbi). Kukhona futhi intshisekelo kuma- ejenti okufunda okuqinisayo ahlala evumelana nedatha entsha yemakethe ngesikhathi sangempela, okungase kubhekane nezimo ezishintshayo kangcono kunamamodeli aqeqeshwe ngokungaguquki. Ngaphezu kwalokho, ukuhlanganisa i-AI namasu avela ezimalini zokuziphatha noma ekuhlaziyweni kwenethiwekhi kungaveza amamodeli acebile okuguquguquka kwemakethe. Noma kunjalo, ngisho ne-AI yesikhathi esizayo ethuthuke kakhulu izosebenza ngaphakathi kwemingcele yamathuba nokungaqiniseki.

Ngamafuphi, umbuzo othi “Ingabe i-AI ingabikezela imakethe yamasheya?” awunayo impendulo elula ethi yebo noma cha. Impendulo enembile kakhulu yile: I-AI ingasiza ekubikezeleni imakethe yamasheya, kodwa akuyona into engenaphutha. Inikeza amathuluzi anamandla, uma esetshenziswa ngokuhlakanipha, angathuthukisa amasu okubikezela nokuhweba, kodwa ayisusi ukungaqiniseki okuyisisekelo kwezimakethe. Abatshalizimali kufanele bamukele i-AI ngenxa yamandla ayo - ukucubungula idatha nokuqashelwa kwamaphethini - ngenkathi behlala beqaphela ubuthakathaka bayo. Ngokwenza kanjalo, umuntu angasebenzisa okuhle kakhulu kokubili: ukwahlulela komuntu kanye nobuhlakani bomshini besebenzisana. Imakethe yamasheya kungenzeka ingalokothi ibe yinto ebikezelwayo engu-100%, kodwa ngokulindela okungokoqobo kanye nokusetshenziswa okuhlakaniphile kwe-AI, abahlanganyeli bemakethe bangalwela izinqumo zokutshala imali ezinolwazi olungcono, ezihlelekile endaweni yezezimali ehlala ishintsha.

Amaphepha Amhlophe ongase uthande ukuwafunda ngemva kwalokhu:

🔗 Imisebenzi Engenakushintshwa yi-AI - Futhi Yimiphi Imisebenzi Ezoshintshwa yi-AI?
Thola ukuthi yimiphi imisebenzi evikela ikusasa nokuthi yimiphi esengozini enkulu njengoba i-AI ishintsha ukuqashwa komhlaba wonke.

🔗 Yini i-AI Ekhiqizayo Ongathenjelwa Kuyo Ukuyenza Ngaphandle Kokungenelela Komuntu?
Qonda imingcele yamanje kanye namakhono azimele e-AI ekhiqizayo ezimweni ezisebenzayo.

🔗 Ingasetshenziswa Kanjani I-AI Ekhiqizayo Ekuphepheni Kwe-Cyber?
Funda ukuthi i-AI ivikela kanjani ezinsongweni futhi ithuthukise ukuqina kwe-cyber ngamathuluzi okubikezela nawokuzimela.