Ukubika okuqondiswe kukhompuyutha: Ingabe izintatheli ze-robo ziba ngokwejwayelekile?

• About the Author:
• Igama lomlobi

  I-Quantumrun Foresight
• January 18, 2023

Ngokubika okuqondiswe kukhompuyutha, imisebenzi eminingi ngokujwayelekile eyenziwa abantu, njengokuhlunga okuphakelayo noma ukubhala izindaba eziyisisekelo, manje ingenziwa ngokuzenzakalela ngosizo lwezinqubo zokucubungula ulimi lwemvelo (NLP) kanye nobuhlakani bokwenziwa (AI). Njengoba la mathuluzi eqhubeka nokuthuthukiswa, azovumela amagumbi ezindaba ukuthi aqeqeshe kabusha abasebenzi noma anciphise izindleko zabasebenzi ngokuqhubekayo.

Umongo wokubika oqondiswe kukhompuyutha

Amabhothi aya ngokuya asetshenziswa ukwenza ngokuzenzakalelayo ukukhiqizwa nokusabalalisa izindaba. Ngokwesibonelo, emizuzwini embalwa nje ukuzamazama komhlaba kugadla eCalifornia, intatheli yerobhothi u-Quakebot ingabhala isihloko futhi isithumele kumhleli ongumuntu ku-Los Angeles Times, bese enquma ukuthi uyayishicilela yini indaba. I-MittMedia yaseSweden ikhulise ukufakwa kwezindlu ezithengiswayo kusukela ezindabeni ezimbili ngenyanga kuya kwezingu-2,000, zonke ziqhutshwa irobhothi elibikayo. Lokhu kunyuka kokubukwa kwamakhasi kanye nababhalisile kungabangelwa ukwenyuka kwevolumu yokuqukethwe.

Ama-algorithms, ahlelwe ngokuqondile noma aqeqeshelwa idatha kusetshenziswa ukufunda ngomshini, aba yinkimbinkimbi kakhulu kunangaphambili. Bangakwazi ukubhala izingxoxiswano ngokushesha, babhale imibiko yezindaba, baphenye imikhondo, futhi bahlele izindaba ukuze bagcine abafundi bebambekile. Ushintsho lwakamuva lubone amathuluzi e-AI ebonakala kakhulu ekubikeni nasekuphenyeni izindima. 

Nakuba kubonakala sengathi abantu abaningi abanandaba noma izintatheli zabo zingabantu noma amarobhothi, ochwepheshe baxwayisa ngokuthi njengoba ama-algorithms enza umsebenzi omningi wobuntatheli, amagumbi ezindaba kufanele athuthukise ukuqonda okunamandla kwezinzuzo nobubi bobuchwepheshe. Inhlolovo ye-Reuters Institute emazweni angama-52 ithole ukuthi amaphesenti angama-40 abaholi bezindaba bakholelwa ukuthi i-robo-journalism, lapho i-AI ibhala khona izindaba ngokuzenzakalelayo, izoba inkambiso ebalulekile yezimboni ngo-2022.

Umthelela ophazamisayo

Ukwamukelwa okwandayo kwe-NLP kubuntatheli kungahamba ngezindlela ezimbili: ukugxila kokuqukethwe kokuhlela okunenani eliphakeme noma ukuzenzela ngokuzenzakalelayo cishe zonke izinqubo ukuze kwenziwe lula ukusebenza futhi konge izindleko zabasebenzi. Isibonelo sakamuva i-Radar yase-UK, okuwukuphela kwesikhungo sezindaba sasendaweni esisebenza ngokuzenzakalelayo emhlabeni wonke eyethulwa ngo-2018. Le nkampani yezindaba yasekelwa Isikhwama Sezindaba Zedijithali Se-Google. 

Le nkampani inethimba lezintatheli ezinhlanu ezifake ama-athikili angaphezu kuka-400,000 kusukela i-Radar yashintshela kumodeli yokubhalisa. Okwenza i-ejensi ithakazelise ukuthi isebenzisa amasethi edatha atholakala esidlangalaleni futhi iwaphakela ama-algorithms esifanekiso ukuze idale izindaba ezihlukene zezindaba ezinama-engeli ahlukahlukene zisebenzisa lezi zibalo. Umbono awugxili ebuntathelini obuphenyayo kodwa ugxile ekudaleni izindaba zansuku zonke ezisekelwe ezinombolweni. 

Le modeli yebhizinisi ingase ithole ukudonseka okwengeziwe njengoba izinkampani eziningi zibona ukubaluleka kokudala ama-algorithms we-AI wangokwezifiso ukuze kukhuliswe ukukhiqizwa kokuqukethwe. Kodwa-ke, izinhlangano zezindaba zibhekana nenkinga yokuziphatha lapho zilungisa amasistimu e-AI ngokwezinjongo zobuntatheli. Isibonelo, abahleli bangakwazi kanjani ukubona ukuchema ku-algorithm? Lingalungiswa kanjani iphutha le-natural language generator (NLG)? Ingabe i-AI inembe ngokwanele ukwazi ukuthi yiluphi ulwazi olubalulekile nolusha? Ingabe i-AI kufanele ibe nesibopho sokukhetha ukuthi yikuphi okuqukethwe okudliwa izigidi zabantu?

Ukuphendula le mibuzo kubalulekile ngoba izinkampani zobuchwepheshe, hhayi izintatheli, yizo ezibhekene nokuqamba okusha kule ndawo. I-Tech firm eyaphenduka inkampani yemidiya i-Knowhere News, ngokwesibonelo, iyaqonda ukuthi izinhlelo zayo ze-ML zingaqoqa ulwazi emithonjeni ekhona kuphela, njengokukhishwa kwabezindaba nokuthunyelwe kwezokuxhumana. Ngakho-ke, amaphuzu okubuka anomkhawulo noma achemile ngokwemvelo, futhi ukucabangela kuyaphazamiseka.

Imithelela yokubika okuqondiswe kukhompuyutha

Imithelela ebanzi yokubika okuqondiswe kukhompuyutha ingase ihlanganise: 

• Ezinye iziqalo ezihlinzeka ngezixazululo ze-AI kumagumbi ezindaba, okuhlanganisa ukwakha amakhasi okufika kanye nokufaka izindonga zokukhokha lapho kufanele khona. Ukunciphisa izindleko zalezi zinsiza kungase kwenze izindaba ezizimele, zendawo, noma ze-niche kanye nobuntatheli busebenze ngokwezimali futhi.
• Izinkampani ezinkulu zezindaba ezithuthukisa amasistimu azo azenzakalelayo futhi ziqashisa isevisi kumasayithi wezindaba amancane.
• Ipharadigm ejulile lapho okuqukethwe nezindaba kungokwesikhathi sangempela, kuningi, kumahhala, futhi kuhlanganiswe kuwo wonke ama-engeli. Abathengi bokuqukethwe bazokhungatheka ngokukhetha futhi bancike kokuphakelayo okukhethiwe kwezindaba zabo. 
• Ubuntatheli obuphenyayo buzokhula ngokubaluleka kanye nenani lezezimali njengoba zonke ezinye izinhlobo zokuqukethwe kwezindaba zizothengiswa. Ubuntatheli bangempela bungabona ukuvuselelwa.  
• Ukwanda kokungathembani kubafundi abakhetha ubuntatheli bendabuko.
• Sabisa abalingisi abafaka imikhankaso yokulahla ulwazi kanye nenkulumo-ze kuma-algorithms ngaphandle kokuthi abahleli bazi ngakho.
• Ukwandisa isidingo somphakathi (noma umthetho) sokuthi izinhlangano zezindaba zibe phambili endleleni ezisebenzisa ngayo i-AI/ML ezinhlelweni zazo futhi zinikeze izintatheli ze-AI imigqa ecacile ecacile.

Imibuzo ongaphawula ngayo

• Ucabanga ukuthi i-robo-journalism izophinde iguqule kanjani indlela abantu abadla ngayo izindaba?
• Yiziphi ezinye zezingosi zakho zezindaba ezethembekile, futhi ngani?