Ukukhathala Kwe-AI Kuyinto Ekhona. Nansi Okukhona Ebhakabhakeni Lakho.

Kukhona okushintshile ngo-2024. Inani labantu abachaza uhlobo oluthile lokukhathala kwengqondo — hhayi ukushoda amandla, hhayi ukugcwala kolwazi, kodwa okuthile okuphakathi kwalezi zimbili — kwanda kakhulu. Incazelo iyafana: umuzwa wokugcwala kwengqondo, ubunzima bokucabanga ngokuzimela, ukuncipha kokubekezelela okungaqinisekile, kanye nokuncika okungakhululekile kumathuluzi e-AI okwenza ukuhamba kokucabanga kube nzima, hhayi lula.

Lokhu akusikho ukwesaba ubuchwepheshe noma ukukhumbula. Kubonisa izimo zangempela ze-neurological ezilindelekile uma sithatha ukuthi siyini esikwaziyo mayelana nokuthi ubuchopho b processing ulwazi, bugcina ukusebenza kahle kwengqondo, futhi baphendula ekusizeni okukhona njalo. Nansi okwenzeka — nokuthi ubufakazi bukhombisa kanjani ukuhamba kulokhu.

Imvelo entsha yokucabanga

Ngokuphathelene no-2025, abasebenzi bezolimo emikhakheni eminingi baxhumana nezinsiza ze-AI — i-ChatGPT, i-Copilot, i-Gemini, i-Claude, kanye nezinsiza ezikhethekile eziningi — izikhathi eziningi ngosuku. Umsebenzisi ojwayelekile uthumela imibuzo eminingi nsuku zonke. Okuqukethwe okwenziwe yi-AI manje sekubamba ingxenye enkulu yemibhalo ye-imeyili, izifinyezo zemihlangano, ikhodi, imibhalo ebhalwe phansi, nezindlela zokusesha.

Lezi yizimo ezintsha zengqondo. Okokuqala emlandweni, ingxenye enkulu yomsebenzi wengqondo obudinga umzamo omkhulu — ukuhlela, ukufingqa, ukufuna, ukuhlela, ukulungisa amaphutha — ingadluliselwa cishe ngokushesha. Umbuzo uthi lokhu kwenza kanjani ubuchopho bethu ngokuhamba kwesikhathi.

Impendulo, ngokusekelwe kulokho esikukwazi mayelana nomsebenzi wengqondo nokukhathala kwengqondo, iyinkimbinkimbi kunalokho okukhuluma kahle (i-AI ikhulisa amandla abantu) noma okukhuluma kabi (i-AI iyancipha amandla abantu). Kokubili kuyenzeka, ngezindlela ezihlukene, ngezinhlobo ezihlukene zokucabanga.

Kungani amathuluzi e-AI engakhuphula ukukhathala kwengqondo esikhundleni sokukwehlisa

Umqondo ojwayelekile ukuthi ukususa umsebenzi ku-AI kunciphisa umthwalo wokucabanga. Kwezinye izindlela, kuyenzeka. Kodwa ucwaningo ngombono womthwalo wokucabanga (Sweller, 1988) luhlukanisa phakathi komthwalo <em>ongaphakathi</em> (ubunzima bomsebenzi uqobo), umthwalo <em>ongaphandle</em> (izidingo zokucabanga ezidalwa indlela umsebenzi ovezwa ngayo), kanye nomthwalo <em>ofanele</em> ( umzamo wokucabanga odala ukuqonda nokwazi).

Amathuluzi e-AI ngokuyinhloko anciphisa umthwalo ongaphakathi nowofanele — abhekana nomsebenzi obungathatha umzamo futhi uveze ukufunda. Kodwa avame <em>ukwandisa</em> umthwalo ongaphandle: ukuhlola umphumela we-AI ukuze uqinisekise ukunembile, ukuhlanganisa izingxenye eziningi ezikhiqizwe yi-AI zibe nomsebenzi ohlangene, ukuphatha umongo phakathi kwamathuluzi amaningi, ukukhetha okungathunyelwa nokuthi yini okufanele yenziwe mathupha, nokugcina ukubhekwa kwezinqubo ze-AI. Lezi yizidingo ezintsha zokucabanga ezazingekho ngaphambili.

Umphumela kubantu abaningi uyiphazili: ukukhiqiza okuningi, ukukhathala kokucabanga okuningi. Umsebenzi uzwakala ulula ezikhathini ezithile kodwa ukhathaza kakhulu ngosuku. Lokhu kuhambisana nocwaningo ngokuqonda <em>ukukhathala kokukhetha</em> — izindleko ezihlanganisiwe zokwenza izinqumo ezincane eziningi zinciphisa izinsiza ezifanayo ze-prefrontal njengokwenza ezimbalwa ezinkulu.

Ukusebenza okusekelwe ku-AI kukhuphula kakhulu inani lezi zinqumo ezincane: yamukela lesi siphakamiso, phika lesi, lungisa le ndima, buza kabusha le mpendulo. Izinqumo zenzeka ngokushesha, kodwa umthwalo ophakathi kwezinkulungwane zezi zinqumo ezincane zansuku zonke unzima.

I-paradox yokwenza izinto ngokuzenzakalelayo kanye nokwehla kwamakhono

Kukhona umphumela owaziwayo ocwaningweni lwe-ergonomics kanye nezici zabantu obizwa nge-<em>automation paradox</em> noma <em>izinkinga zokwenza izinto ngokuzenzakalelayo</em> (Bainbridge, 1983): lapho ukwenza izinto ngokuzenzakalelayo kuthatha umsebenzi ojwayelekile, abaphathi abantu bahluleka ukujwayela amakhono adingekayo ezimweni ezingajwayelekile. Abaphayona abathembela ku-autopilot ngesikhathi esiningi sokundiza baveza amakhono okuphatha ngesandla aphansi. Abahloli bezithombe abawasebenzisayo amathuluzi e-AI wokuxilonga baveza izinga lokuthola eliphansi uma i-AI iphutha okuthile.

Le ndlela efanayo iyasebenza emsebenzini wezobuchopho. Ukubhala, ukucabanga ngezinkinga, ukuhlela, nokuhlanganisa okwakhiwe yiziphiwo ezithuthukiswa ngokuzijwayeza futhi zinciphisa uma zingasetshenziswa. Lapho i-AI iphethe umqulu wokuqala, ukuhlanganiswa, noma ukuhlela okuhlelekile, lezi zindlela zobuchopho zithola ukuncipha. Ukuhlupheka akubukeki kakhulu — akusho ukuthi abantu bayakhohlwa indlela yokucabanga — kodwa ukuncipha kancane kwezifundo ezijwayelekile kukhiqiza ushintsho olukhulu emandleni obuchopho nasekuqiniseni ngokuhamba kwesikhathi.

Lokhu kuhlukile ekubeni nje ungajwayele ukubhala ngempensela. Ukubhala — empeleni ukuhlela, hhayi ukuhlela noma ukuhlela — kubandakanya imemori esebenzayo, ukuhlela imibono, nokucabanga okuhlola ngendlela yokuthi ukubuyekeza nokuhlela okukhishwe yi-AI akukwazi ukukwenza ngokuphelele. Umsebenzi wezobuchopho wokuhlela uyakha; umsebenzi wezobuchopho wokuhlela uyahlola. Bobabili banezindleko; abahlanganyeli.

Iphutha lokuhlukana kokunaka

Ngaphandle kokukhathazeka kokucabanga nokwehla kwamakhono, amathuluzi e-AI engeza enye ingxenye yokuphazamiseka kokugxila endaweni esivele igcwele ukuphazamiseka. Yonke i-interface yokuxoxa, umsizi wokubhala, kanye nezinsiza zokubhala iyinsiza yokuphazamiseka engaba khona — umbuzo wokwakha, umphumela wokuhlola, ukulandela ukuze uthuthukise.

I-prefrontal cortex, ephatha ukugxila okuqhubekayo nemisebenzi yokuphatha, ayihlukanisi phakathi kokuphazamiseka okuveliswa abantu kanye nokuphazamiseka okuveliswa yi-AI. Ucwaningo olwenziwa ngezindleko zokushintsha imisebenzi (Rubinstein et al., 2001) lukhombisa ukuthi ukushintsha phakathi kwemisebenzi — ngisho nangokwesikhashana — kubeka umthwalo wokucabanga ophindaphindayo ngosuku. Ukwengeza amathuluzi amaningi e-AI endaweni yokusebenza esivele iphukile kwenza kube kubi lokhu ngaphandle kokuthi abaningi baqaphele imvelaphi yokukhathala kwabo.

I-ironiy ukuthi amathuluzi e-AI avame ukwamukelwa ikakhulukazi ukuze kuncishiswe umthwalo wokucabanga. Kodwa ukwengeza amathuluzi amasha, ama-interface amasha, kanye nezindawo zokwenza izinqumo ezisha kuvame ukwandisa umthwalo wokucabanga jikelele kunokuwunciphisa — okungenani kuze kube yilapho la mathuluzi ehlanganiswe ngokwanele ukuze aphathwe ngokuzenzakalelayo kunokucabangela.

Yini okufanele uyenze ngempela ngayo

<strong>Hlukanisa phakathi kokunciphisa nokukhipha.</strong> Akukho ukusiza kwe-AI okunezimpempe ezifanayo. Ukusebenzisa i-AI ukuphatha imisebenzi enenani eliphansi (ukuhlela, izifanekiso, ukufuna okuphindaphindiwe) kunciphisa umthwalo ngaphandle kokulahlekelwa kakhulu ngamakhono. Ukusebenzisa i-AI ukuphatha ukucabanga ofuna ukuhlala unakho — ukubhala okuyinkimbinkimbi, ukuhlaziywa okucacile, ukuxazulula izinkinga ngokwakha — kunciphisa umthwalo ngentengo yokulahlekelwa kwamakhono. Ukuqonda kahle ukuthi iyiphi isigaba esithile sokusetshenziswa sihamba kanjani, kuyisiqalo.

<strong>Vikela isikhathi sokucabanga ngaphandle kosizo.</strong> Amakhono okucabanga aphakeme kakhulu athinteka kakhulu ekudlulisweni kwe-AI yilezi ezibalulekile kakhulu ukuziqinisekisa: ukucabanga ngokuzimela, ukubhala okuqhubekayo, ukuhlela okuyinkimbinkimbi. Ukuziqeqesha ngalezi zikhono ngaphandle kosizo lwe-AI — ngezikhathi ezichaziwe, emisebenzini ebalulekile — akusikho ukwesaba ubuchwepheshe; kuyisizathu esifanayo sokuthi udokotela wezinso azilolonge izindlela zokusebenza ngenkathi amathuluzi e-robotic ethola.

<strong>Nciphisa inani lamathuluzi e-AI asetshenziswa ngokuqhubekayo.</strong> Izindleko zokuphatha izixhumi eziyisihlanu ezahlukene ze-AI ziphakeme kunezokuphatha eyodwa ekhethiwe kahle. Ukukhula kwamathuluzi e-AI kudala inkinga efanayo yokuphazamiseka kokunaka njengokukhula kwezinhlelo zokusebenza nezaziso. Ukuhlanganiswa kwehlisa umthwalo wokucabanga ongeyona eyodwa.

<strong>Sebenzisa izimiso ezifanayo njengokuphila kahle kwedijithali.</strong> Ukukhathala kwe-AI kuyifomu ethile yenkinga enkulu i-Unwire ebhekana nayo: izindawo zokucabanga ezidinga okuningi kunokuhlinzeka, ukuphazamiseka kokunaka okunciphisa ngaphandle kokubuyisela, kanye nokukhuthazwa okukhona njalo okwenza ukuba ukwehlukanisa kube nzima. Izinyathelo zihlala zifanayo: izikhathi ezichaziwe zokuphumula, ukwakhiwa kwemvelo okunciphisa ukuxhumana kwe-AI, kanye nesikhathi sokubuyisela esijwayelekile lapho ubuchopho bungaholelwa.

Umbono wesikhathi eside

Ukuhlanganiswa kwe-AI emsebenzini wolwazi akuzukuphenduka. Umbuzo uwukuthi abantu bazohamba kanjani endaweni ethile, eyenzelwe ukugxila kakhulu nokunciphisa ukuhamba — nokuthi lokho kwenza kanjani empilweni yokucabanga eminyakeni esikhathini eside kunokuhamba kwezinsuku.

Ucwaningo olukhuluma ngempilo yedijithali lukhombisa ukuthi abantu abaphumelelayo ezindaweni eziphezulu zobuchwepheshe akusikho labo abasebenzisa amathuluzi amaningi noma ambalwa, kodwa labo abagcina ukulawula ngokucophelela izimo zokugxila kwabo: lapho betholakala, lapho bengatholakali, lokho abakuthumelayo, nokuthi yini abayivikela.

Lokhu kukhuthazela kunzima ukukugcina endaweni egcwele i-AI kunokuthi endaweni egcwele ama-smartphone kuphela. Kodwa umgomo oyisisekelo — vikela amandla akho okucabanga ngokuzimela, buyisela isisekelo sebhayisikobho ngokuphumula okuyiqiniso, futhi udale indawo yakho kunokuthi uphendule kuyo — kuhlala kufana.

Sources

1. Sweller, J. (1988). Cognitive load during problem solving: effects on learning. Cognitive Science, 12(2), 257–285.
2. Bainbridge, L. (1983). Ironies of automation. Automatica, 19(6), 775–779.
3. Rubinstein, J.S., et al. (2001). Executive control of cognitive processes in task switching. Journal of Experimental Psychology: Human Perception and Performance, 27(4), 763–797.
4. Wiehler, A., et al. (2022). A neuro-metabolic account of why daylong cognitive work alters the control of economic decisions. Current Biology, 32(16), 3564–3575.
5. Baumeister, R.F., et al. (1998). Ego depletion: is the active self a limited resource? Journal of Personality and Social Psychology, 74(5), 1252–1265.