Lapho abantu bekhuluma nge-AI kulezi zinsuku, ingxoxo ihlale igxumela kuma-chatbots azwakala njengomuntu, amanethiwekhi amakhulu e-neural ancipha idatha, noma lezo zinhlelo zokuqaphela izithombe ezibona amakati kangcono kunabantu abakhathele. Kodwa esikhathini eside ngaphambi kwalowo mdumo, kwakukhona i-Symbolic AI . Futhi ngokumangazayo - isekhona, isasebenziseka. Ngokuyisisekelo imayelana nokufundisa amakhompuyutha ukucabanga njengoba kwenza abantu: ukusebenzisa izimpawu, ukuqonda, nemithetho . Isidala? Kungenzeka. Kepha emhlabeni ogcwele “ibhokisi elimnyama” AI, ukucaca kwe-Symbolic AI kuzwakala kuqabula [1].
Izindatshana ongathanda ukuzifunda ngemva kwalesi:
🔗 Yini umqeqeshi we-AI
Ichaza indima nezibopho zabaqeqeshi besimanje be-AI.
🔗 Ingabe isayensi yedatha izothathelwa indawo yi-AI
Ihlola ukuthi ukuthuthuka kwe-AI kuyayisongela yini imisebenzi yesayensi yedatha.
🔗 I-AI ilutholaphi ulwazi lwayo
Ihlukanisa imithombo ye-AI esetshenziswa amamodeli ukuze afunde futhi azivumelanise nezimo.
Izisekelo ze-AI ezingokomfanekiso✨
Nansi isivumelwano: I-symbolic AI yakhelwe phezu kokucaca . Ungakwazi ukulandelela i-logic, uhlole imithetho, futhi ubone ukuthi kungani umshini usho lokho okwenzile. Qhathanisa lokho nenethi ye-neural evele ikhafula impendulo - kufana nokubuza osemusha ukuthi "kungani?" nokuthola amahlombe. Izimiso ezingokomfanekiso, ngokuphambene, zizothi: “Ngenxa yokuthi u-A no-B basho u-C, ngakho-ke u-C.” Lelo khono lokuzichaza liyinguquko-midlalo yezinto ezisezingeni eliphezulu (umuthi, ezezimali, ngisho nenkantolo) lapho othile ehlala ecela ubufakazi [5].
Indaba encane: ithimba lokuthobela umthetho ebhange elikhulu elifake izinqubomgomo zonswinyo libe yinjini yemithetho. Izinto ezifana nokuthi: “uma i-original_country ∈ {X} kanye ne- missing_beneficiary_info → zikhula.” Umphumela? Icala ngalinye elihlatshwe umkhosi lifike nochungechunge lokucabanga olulandelekayo, olufundeka umuntu. Abacwaningi mabhuku bayithandile . Lokho amandla amakhulu e-Symbolic AI - asobala, ukucabanga okuhlolekayo .
Ithebula Lokuqhathanisa Ngokushesha 📊
| Ithuluzi / Indlela | Ubani Oyisebenzisayo | I-Cost Range | Kungani Isebenza (noma ingasebenzi) |
|---|---|---|---|
| Amasistimu Ochwepheshe 🧠 | Odokotela, onjiniyela | Ukusetha okumba eqolo | Ukucabanga okusekelwe emthethweni okucace kakhulu, kodwa kuhlakaniphe [1] |
| Amagrafu olwazi 🌐 | Izinjini zokusesha, idatha | Izindleko ezihlanganisiwe | Ixhuma amabhizinisi + nobudlelwano esikalini [3] |
| Ama-Chatbots asekelwe emthethweni 💬 | Insizakalo yekhasimende | Okuphansi-okumaphakathi | Ngokushesha ukwakha; kodwa umehluko? hhay kangako |
| I-Neuro-Symbolic AI ⚡ | Abacwaningi, iziqalo | Phezulu phezulu | I-Logic + ML = iphethini echazekayo [4] |
Isebenza Kanjani I-AI Engokomfanekiso (Ezingokoqobo) 🛠️
Emgogodleni wayo, i-Symbolic AI yizinto ezimbili nje: izimpawu (imiqondo) kanye nemithetho (ukuthi leyo miqondo ixhuma kanjani). Isibonelo:
-
Izimpawu:
Inja,Isilwane,I-HasTail -
Umthetho: Uma u-X eyinja → u-X uyisilwane.
Kusuka lapha, ungaqala ukwakha amaketanga engqondo - njengezingcezu ze-LEGO zedijithali. Amasistimu ochwepheshe bakudala aze agcina amaqiniso ngokuphindwe kathathu (isibaluli–into–inani) futhi asebenzisa umhumushi wesimiso esiqondiswe kumgomo ukuze afakazele imibuzo isinyathelo ngesinyathelo [1].
Izibonelo Zempilo Yangempela Ye-Symbolic AI 🌍
-
I-MYCIN - uhlelo lochwepheshe bezokwelapha lwezifo ezithathelwanayo. Okusekelwe emithethweni, okuvumelana nezincazelo [1].
-
I-DENDRAL - i-chemistry yokuqala ye-AI eqagela izakhiwo zamangqamuzana kusuka kudatha ye-spectrometry [2].
-
I-Google Knowledge Graph - izinhlangano zamamephu (abantu, izindawo, izinto) + nobudlelwano babo ukuze baphendule imibuzo “yezinto, hhayi izintambo” [3].
-
Ama-bots asekelwe kumthetho - ukugeleza okubhaliwe kokusekelwa kwamakhasimende; iqinile ukuze ihambisane, ibuthakathaka engxoxweni evulekile ye-chit.
Kungani i-Symbolic AI Yakhubeka (kodwa Ayifanga) 📉➡️📈
Nakhu lapho i-Symbolic AI ikhuphuka khona: umhlaba wangempela ongcolile, ongaphelele, ophikisanayo. Ukugcina isisekelo esikhulu somthetho kuyakhathaza, futhi imithetho eqinile ingasebenza ibhaluni ize iphuke.
Nokho - akuzange kuhambe ngokugcwele. Faka i-AI ye-neuro-symbolic : hlanganisa amanethi e-neural (ahle ngombono) nomqondo ongokomfanekiso (okahle ekucabangeni). Kucabange njengeqembu elidluliswayo: ingxenye ye-neural ibala uphawu lokuma, bese ingxenye engokomfanekiso ithola ukuthi isho ukuthini ngaphansi komthetho womgwaqo. Leyo nhlanganisela ithembisa amasistimu ahlakaniphile futhi achazekayo [4][5].
Amandla e-Symbolic AI 💡
-
I-logic esobala : ungalandela zonke izinyathelo [1][5].
-
Imithethonqubo ilungele : amamephu ahlanzekile kuzinqubomgomo nemithetho yezomthetho [5].
-
Ukugcinwa kwe-modular : ungashintsha umthetho owodwa ngaphandle kokuqeqesha kabusha yonke imodeli ye-monster [1].
Ubuthakathaka be-Symbolic AI ⚠️
-
Kubi ekubonweni : izithombe, okulalelwayo, umbhalo ongcolile - amanetha e-neural abusa lapha.
-
Izinhlungu zokukala : ukukhipha nokuvuselela imithetho yochwepheshe kuyakhathaza [2].
-
Ukuqina : imithetho yephula ngaphandle kwendawo yayo; ukungaqiniseki kunzima ukukuthwebula (yize amanye amasistimu agqekeza ukulungisa ingxenye) [1].
Indlela Engaphambili Ye-Symbolic AI 🚀
Ikusasa cishe aliyona i-neural emsulwa engokomfanekiso noma emsulwa. Kuyihybrid. Cabanga nje:
-
I-Neural → ikhipha amaphethini kumaphikseli aluhlaza/umbhalo/umsindo.
-
I-Neuro-symbolic → iphakamisa amaphethini abe yimiqondo ehlelekile.
-
I-Symbolic → isebenzisa imithetho, imingcele, bese - okubalulekile - ichaza .
Yilapho imishini iqala khona ukufana nokucabanga komuntu: bheka, isakhiwo, thethelela [4][5].
Ukusonga 📝
Ngakho-ke, i-Symbolic AI: iqhutshwa ngokunengqondo, isekelwe emthethweni, ilungele incazelo. Ayinabukhazikhazi, kodwa ibethelela okuthile amanetha ajulile namanje angakwazi: ukucabanga okucacile, okufundekayo . Ukubheja okuhlakaniphile? Amasistimu aboleka kuwo womabili amakamu - amanetha e-neural okubona nesikali, afanekisela ukucabanga nokwethemba [4][5].
Incazelo ye-Meta: I-AI yophawu ichazwe - amasistimu asekelwe kumthetho, amandla/ubuthakathaka, nokuthi kungani i-neuro-symbolic (logic + ML) iyindlela eya phambili.
Ama-Hashtag:
#ArtificialIntelligence 🤖 #SymbolicAI 🧩 #MachineLearning #NeuroSymbolicAI ⚡ #TechExplained #KnowledgeRepresentation #AIInsights #FutureOfAI
Izithenjwa
[1] Buchanan, BG, & Shortliffe, EH Amasistimu Ochwepheshe Asekelwe Kumthetho: Ukuhlolwa kwe-MYCIN Kwephrojekthi Yezinhlelo Ze-Stanford Heuristic , Ch. 15. PDF
[2] Lindsay, RK, Buchanan, BG, Feigenbaum, EA, & Lederberg, J. “DENDRAL: isifundo esiyisibonelo sohlelo lokuqala lochwepheshe lokwakhiwa kwe-hypothesis yesayensi.” I-Artificial Intelligence 61 (1993): 209–261. PDF
[3] Google. "Sethula Igrafu Yolwazi: izinto, hhayi izintambo." I-Google Blog Esemthethweni (Meyi 16, 2012). Isixhumanisi
[4] Monroe, D. “Neurosymbolic AI.” Ukuxhumana kwe-ACM (Oct. 2022). I-DOI
[5] Sahoh, B., et al. "Indima ye-Artificial Intelligence echazekayo ekuthathweni kwezinqumo eziphezulu: isibuyekezo." Amaphethini (2023). I-PubMed Central. Isixhumanisi