Unxibelelwano lweBrain-Computer (BCI): Ukuya ekudibaneni kwabantu ne-AI 

Uvavanyo oluqhubekayo lweklinikhi lweBrain-Computer Interfaces (BCIs) olufana ne-Neuralink's “Telepathy” implant lubandakanya ukuseka amakhonkco onxibelelwano phakathi kweengqondo zabathathi-nxaxheba abaneemfuno zonyango ezingafezekiswanga ngenxa yokonakala kweendawo zokunxibelelana zebhayoloji kwiimeko ezifana nokwenzakala komqolo, istroke, okanye i-amyotrophic lateral sclerosis (i-ALS)) kunye namaqonga e-AI. I-BCI implant ithatha umsebenzi weendawo zokujonga izinto ezonakeleyo, kwaye abathathi-nxaxheba kuvavanyo bayakwazi ukusebenzisa iifowuni, iikhompyutha, iilaptop, imidlalo, kunye neengalo zerobhothi ngokucinga kuphela. Le nkqubela ibonisa ukuba, kungekudala, kunokwenzeka ukuseka unxibelelwano olukhawulezayo phakathi kwengqondo kunye neeplatifomu ze-AI ezidlula iindawo zethu zokujonga izinto eziphilayo ezicothayo kakhulu kwaye zoyise imida ye-bandwidth ukudibanisa i-AI kwinqanaba lethu le-compute lesithathu. Amakhonkco e-neural ane-bandwidth ephezulu aya kusebenza njengebhulorho, edibanisa ngempumelelo ingqondo ne-AI. Abantu baya kuba zii-cyborgs (ii-cybernetic organisms). Ukudibana kuya kwenza ukuba ezi zimbini zizuze omnye komnye. Ingqondo iya kufumana amandla e-computing angaphezu kwawemvelo e-AI ngaloo ndlela inciphise umngcipheko wokuba abantu baphelelwe lixesha xa bejongene nezidalwa zedijithali ezikrelekrele kakhulu. I-symbiosis yengqondo yomntu-AI iya kuba yimpendulo kumngcipheko wobomi kuluntu obangelwa yi-AI ekrelekrele kakhulu.       

Inkqubo yobukrelekrele bokwenziwa (AI) yimodeli yolwimi (LM) eyenza uqikelelo olunokwenzeka lwegama elilandelayo kulwimi lwendalo olunikwe elo (ezingaphambili). Imodeli iqeqeshwa kwangaphambili ngedatha ukuze iqikelele okuza kulandela kwizivakalisi xa icelwa. Ngokwenza njalo, imodeli ilinganisa umsebenzi wobukrelekrele bendalo.   

Iindlela ezindala zokuqiqa ezisekelwe kwi-AI. Zazisekelwe kwingcamango yokuba intsingiselo yobukrelekrele bomntu kukuqiqa okanye ingqiqo. Ngokwale ndlela yokufuzisela, intsingiselo yegama yindlela elihambelana ngayo namanye amagama. Ukuqonda isivakalisi kuthetha ukuguqulela isivakalisi kulwimi oluthile lwangaphakathi lokufuzisela. Emva koko, umntu usebenzisa imithetho kwiintetho zokufuzisela ukuze afumane iintetho ezintsha. Iinkqubo zokuqala ezikrelekrele ezisekelwe kule ngcamango azizange zisebenze kakuhle, kwaye akukho nkqubela ibalulekileyo enokwenziwa kwicandelo nangona i-AI yaqala kwangeminyaka yoo-1950.  

Kuye kwenziwa inkqubela enkulu kwi-AI kwiminyaka yakutshanje. Kuye kwavela iindlela ezintsha ze-AI ezisebenza kakuhle kakhulu. Zininzi izinto ezisebenzisanayo ukuze oko kwenzeke, enye yazo kukugxininisa kwindlela yebhayoloji okanye yengqondo yokuqonda ingqondo yomntu kunye nendlela esebenza ngayo ingqondo yomntu. Ngokwendlela yebhayoloji, intsingiselo yegama yiseti yeempawu okanye iimpawu kwaye ukuqonda kuthetha ukuguqula uphawu ngalunye lwegama lube ziimpawu ezininzi. Iifom ezintsha ze-AI zidibanisa ezi ndlela zimbini. Iguqula igama ngalinye libe ziimpawu ezinkulu. Ukusebenzisana phakathi kweempawu zamagama ahlukeneyo kwenza kube lula ukuqikelela iimpawu zegama elilandelayo nto leyo eyenza kube lula ukuqikelela igama elilandelayo ngenxa yeempawu zalo.  

I-AI entsha yenza umzekelo wokuqonda komntu (endaweni yokuqiqa). Isekelwe kwinethiwekhi ye-neural kwaye icwangcisa idatha ngendlela efana nengqondo yomntu. Imodeli yolwimi lwenethiwekhi ye-neural enkulu yenza imisebenzi eyahlukeneyo yokucubungula ulwimi lwendalo ngendlela efanelekileyo. Iimodeli ezibalulekileyo zeLarge Language (LLM) zangoku ezifana ne-xAI's Grok, i-Gemini kaGoogle, i-Anthropic's Claude, i-OpenAI's ChatGPT, i-High-Flyer's DeepSeek kunye nezinye zinamandla amakhulu ekhompyutha. Ziqeqeshwe kakuhle kwaye zisebenza kakuhle kakhulu. Amandla azo ekhompyutha angenakuthelekiswa nanto achaphazele iindawo ezininzi. Kukho ingxelo yokuba i-Anthropic's Claude isetyenziselwa uhlalutyo, ukuchonga iipateni, ukucwangcisa, ukulinganisa, imidlalo yemfazwe kwimfazwe eqhubekayo ngoku kummandla we-Middle East.   

Itekhnoloji yeBrain-Computer Interfaces (BCI) yenye yeendawo ezinjalo ezizuze kakhulu kuphuhliso lwakutshanje kwi-AI. Le teknoloji ayintsha kodwa amandla amakhulu ekhompyutha ee-LLM zamva nje enze lula ukuqondwa nokucutshungulwa kwemiqondiso ye-neural. Ngenxa yoko, izixhobo ezininzi ze-BCI ngoku zifikelele kwinqanaba lovavanyo lweklinikhi.  

I-Neuralink, omnye wabadlali ababalulekileyo kweli candelo, iphuhlisa isixhobo sokufakelwa ingqondo, isixhobo sokunxibelelana kwengqondo nekhompyutha (i-BCI) esibizwa ngokuba yi-“Telepathy” esiza kuphucula ukuzimela nokuzimela kwabantu abaneemeko ezibuthathaka ezifana nokwenzakala komqolo, isifo sohlangothi, i-ALS, njl. Iya kwenza abantu abanjalo bakwazi ukulawula ngokuthe ngqo iikhompyutha, iifowuni, kunye nezixhobo zokuncedisa ezifana namalungu erobhothi besebenzisa iingcinga zabo zodwa (i-telepathy, kwisayensi yokuziphatha, ibhekisa kwisiganeko se-parapsychological esibandakanya unxibelelwano oluthe ngqo lwengcinga ukusuka engqondweni yomntu omnye ukuya engqondweni yomnye umntu ngaphandle kokusebenzisa ijelo leemvakalelo eliqhelekileyo kunye nazo naziphi na iimpawu ezaziwayo). Esi sixhobo se-BCI okwangoku sivavanywa kathathu kwangethuba. Ngelixa uPhononongo lwe-PRIME olubandakanya abathathi-nxaxheba abali-15 luvavanya ulawulo lwe-neuronal lwezixhobo zangaphandle, uPhononongo lwe-CONVOY kubathathi-nxaxheba abathathu luphanda ulawulo lwezixhobo zokuncedisa kwaye uPhononongo lwe-VOICE kubathathi-nxaxheba aba-6 luhlola ukubuyiselwa kwefowuni ekhumbuza indlela uStephen Hawking aboniswe ngayo ukuba unxibelelana kwi-sitcom kamabonakude ethi “Big Bang Theory”. Esinye isixhobo sokufakelwa kwengqondo sikaNeuralink esithi “Blindsight”, isixhobo sokubuyisela umbono sisendleleni yovavanyo lweklinikhi olulindele imvume yomthetho. 

Izixhobo zonyango ze-BCI eziphuhliswa yiNeuralink zithatha indawo ye-neuronal interfaces eyonakeleyo kwaye zibuyisele ukusebenzisana kwendalo nokubonakalayo nehlabathi ledijithali nelezinto ezibonakalayo kwabo baneemfuno zonyango ezingafezekiswanga. Isixhobo seTelepathy sithatha isignali yomyalelo evela engqondweni size siyithumele kwizixhobo zangaphandle ezifana nekhompyutha, ifowuni okanye isixhobo sokuncedisa ukwenza umsebenzi. Kwelinye icala, isixhobo seBlindsight siya kucubungula imiqondiso yeemvakalelo eqokelelwe kwindawo yangaphandle ukuze ibonwe ngamehlo yingqondo. Kwimeko enjalo, imiqondiso evela kwindawo yangaphandle iya kuguqulwa ibe yimiqondiso ye-neural ngoncedo lwe-AI kwaye ifakwe kwi-cortex ebonakalayo ukuze ibonwe ngaphaya kwe-interface yeemvakalelo eyonakeleyo. Ukuchonga kunye nokucubungula imiqondiso kuye kwenzeka ngenxa yee-LLM zanamhlanje. Impumelelo ikwabangelwa kukufakelwa kwe-1024-channel okuphucule kakhulu izinga lokudluliselwa kwedatha ukusuka ebuchotsheni ukuya kwikhompyutha. Nangona kusesekwinqanaba lovavanyo lweklinikhi, ezi zifakelo ze-BCI ziya kuphucula kakhulu umgangatho wobomi babantu abachaphazelekayo xa zithengiswa kungekudala. Nangona kunjalo, kukho okungakumbi kwibali lokuqhubela phambili kwitekhnoloji ye-BCI.    

Kwizilingo zeklinikhi ezikhankanyiweyo apha ngasentla, i-AI isetyenziselwa ukucacisa nokucubungula imiqondiso ye-neural eqokelelwe ziimplants ezingqondweni zabantu abaneemfuno ezingafezekiswanga apho ingqondo idlula i-interfaces eyonakeleyo yebhayoloji kwaye inxibelelana ngqo nekhompyutha yangaphandle. Ngaba umntu ophilileyo angasebenzisa amandla amakhulu ekhompyutha eqonga le-AI ngendlela efanayo ukuphucula ukusebenza kakuhle kunye nokusebenza ukuze abe ngumntu onamandla angaphezu kwawemvelo? 

Nantsi isicatshulwa esivela kwinto eyathethwa yingcali yefiziksi uMichio Kaku malunga ne-AI xa wayexoxa ngetekhnoloji yexesha elizayo ngo-2018:  “…Ndicinga ukuba eyona nto iphambili malunga nokuba iirobhothi ziba yingozi nini kukuba xa zizazi kakuhle, mhlawumbi ekupheleni kwenkulungwane. Okwangoku, iirobhothi zethu eziphambili zinobukrelekrele be-cockroach — i-lobotomized cockroach engapheliyo. Kodwa ekugqibeleni iirobhothi zethu ziya kuba krelekrele njengempuku, emva koko zibe krelekrele njengempuku, emva koko zibe nomvundla, emva koko zibe yinja nekati, kwaye ekupheleni kwale nkulungwane, mhlawumbi zibe krelekrele njengenkawu. Ngelo xesha, zinokuba yingozi. Iinkawu ziyazi ukuba ziinkawu. Iinkawu ziyazi ukuba azingobantu. Ngoku, izinja ziyadideka. Izinja azazi ukuba asingobantu. Izinja zicinga ukuba sizizinja kwaye ngenxa yoko ziyasithobela — sizizinja eziphambili, zona zingaphantsi. Ngoko ke ndicinga ukuba, ngelo xesha, kwiminyaka elikhulu ukususela ngoku, ekupheleni kwenkulungwane, sifanele sibeke i-chip engqondweni yazo ukuze sizivale ukuba zineengcinga zokubulala. Leyo yindlela ekhuselekileyo yokusilela, kodwa oko kungokwexeshana kuba ke kwenzeka ntoni xa iirobhothi zikrelekrele kangangokuba zisusa inkqubo ekhuselekileyo yokusilela? Yiyo loo nto Kwakhona kunokwenzeka kwinkulungwane elandelayo, inkulungwane yama-22. Ngelo xesha, ndicinga ukuba sifanele sidibane nabo. Andicingi ukuba oku kuya kwenzeka kule nkulungwane, kodwa ndicinga ukuba kwinkulungwane elandelayo sifanele sidibane nendalo yethu. Kutheni singabi yiHomo superior? Kutheni singasebenzisi ii-exoskeletons ezidalwe ngoku ukuba yiHercules? Lawo ngamandla kathixo. Ngoko ke, ngamanye amazwi, ukhetho endaweni yokulwa neerobhothi, kwinkulungwane ezayo, kukudibana nazo ukuze zibe ngabantu abangaphezulu kwabantu…” - Michio Kaku (2018)Ubuchwephesha bexesha elizayo.

Ekubeni uMichio Kaku enze esi saziso singentla ngo-2018, wathi kwixesha elizayo, “umntu uza kudibana neerobhothi ukuze abe ngumntu onamandla angaphezu kwawemvelo", Itekhnoloji yeBrain-Computer Interfaces (BCI) ibonakala ngathi iqhubela phambili ukuya kuloo ngqikelelo ngenxa yokuqhubela phambili kubuchule bokubala beenkqubo zobukrelekrele bokwenziwa (AI). 

Inkqubo ye-limbic yamandulo yobuchopho bethu (ingqondo yeemvakalelo) ngumthombo wenjongo uninzi lwethu ixesha elininzi. I-cortex yethu yobuchopho (ingqondo yokucinga nokucwangcisa) isebenzisa inani elikhulu le-compute njengomaleko wesibini ukukhonza inkqubo ye-limbic. Ngokwenza njalo, i-cortex yongezwa ngumaleko we-tertiary computing oquka iifowuni, iilaptops, ii-iPad, kunye nezicelo kubandakanya iiplatifomu ze-AI ukuphucula ukusebenza. Ingqondo kweli tyala inxibelelana nomaleko we-tertiary computing ngokusebenzisa ii-interfaces zethu zebhayoloji nokuba ngokuchwetheza okanye ngokuthetha apho izinga lokudluliselwa kwedatha ukusuka kwi-cortex ukuya kumaleko we-tertiary computing liphantsi kakhulu ngoko ke i-bottleneck. Ngaba iingqondo zabantu zinokunxibelelana namaqonga e-AI ngesantya esiphezulu esifana neenkqubo ze-AI computing ezikrelekrele?   

Uqhagamshelo olukhawulezayo oluvumela ukusasazeka kwedatha okuqinisekileyo ngqo kwi-cerebral cortex ukusuka kwi-AI (kwaye ngokuchaseneyo ukusuka kwi-cortex ukuya kwi-AI) kuya kunceda ukudibanisa i-AI ngokufanelekileyo kwi-tertiary compute layer yethu. Yile nto kanye eyenzekayo kwizilingo zeklinikhi ezikhankanyiweyo apha ngasentla - ii-Neuralink's Telepathy implants ziseka uqhagamshelo olukhawulezayo phakathi kwengqondo (yabantu abaneemfuno zonyango ezingafezekiswanga) kunye nekhompyutha edlula kwiindawo ezonakeleyo zebhayoloji ngaloo ndlela idibanisa i-AI kwi-tertiary compute layer yabo. Ngenxa yoko, abathathi-nxaxheba kuvavanyo bayakwazi ukusebenzisa iifowuni kunye neekhompyutha ukukhangela i-intanethi, ukuthumela imiyalezo kunye nokubhala ii-imeyile, ukudlala imidlalo yevidiyo, kunye nokusebenzisa amalungu e-robotic kwimisebenzi ebandakanya ubuchule bezandla ngeengcinga zodwa. Ikhono elitsha liphucula kakhulu umgangatho wobomi babathathi-nxaxheba. Ngokwembono yetekhnoloji, ukudibanisa i-AI kwi-tertiary compute layer yethu ukwandisa umsebenzi ngoqhagamshelo oluphezulu lwe-bandwidth phakathi kwengqondo kunye nekhompyutha (ukutshintsha iindawo zethu ezicothayo zebhayoloji) sisiganeko esibalulekileyo. 

Kakade ke, kukho imeko eqinileyo yokulandela ubuchwepheshe bokuhlangabezana neemfuno zonyango, kodwa kuthekani ngokuhlanganisa i-AI kwi-tertiary compute layer yethu yokwandisa imisebenzi phakathi kwabantu abasempilweni? Le teknoloji ayikude kakhulu; sele ivavanywa ngabantu nangona ijongwa kubantu abaneemfuno zonyango ezingafezekiswanga. Kodwa ngaba iya kuphelela apho?   

Okumangalisayo kukuba, i-AI sele ikwinqanaba lethu le-computing lesithathu kunye nazo zonke ezinye izinto zekhompyutha kwaye yongeza imisebenzi ukuya kwinqanaba apho ii-interfaces zethu zebhayoloji ezicothayo zinokuvumela khona. Sithumela idatha malunga ne-10 ukuya kwi-100 bits ngomzuzwana (bps), umndilili ngaphezulu kweeyure ezingama-24 malunga ne-1 bit ngomzuzwana (bps). Ngoko ke, sisebenzisana namaqonga e-AI ngokusebenzisa ii-interfaces zethu zebhayoloji ezicothayo kakhulu eziyimiqobo kunxibelelwano lwengqondo ne-superintelligent AI. Ngoko ke, kukho ukungafani okukhulu - singathumela malunga ne-10 ukuya kwi-100 bits ngomzuzwana ngelixa ii-AI zangoku zinokucubungula kwaye zikhuphe iitriliyoni zee-bits ngomzuzwana. Oku kuthetha ukuba amandla ethu okunxibelelana nenjongo yethu kwi-AI, kunye nokukwazi kwe-AI ukondla ukuqonda okuntsonkothileyo engqondweni yethu, kuyaphazamiseka yibhayoloji yethu. Ngenxa yoko, ezi zimbini (oko kukuthi, ubuchopho kunye ne-AI) zihlala zingaphandle komnye nomnye. Ngokucacileyo, abantu basengozini yokuba ngabaphelelwe lixesha xa bejongene nee-AI ezikrelekrele. Kukho umngcipheko wokuphila kuluntu. Ngaba i-AI ingamiswa ngenxa yemingcipheko? Kubonakala ngathi akunakwenzeka kuba inesizathu esinamandla kwezoqoqosho kwiinkampani ngokubhekiselele ekunciphiseni iindleko zokusebenza kunye nokuphucula inzuzo. Okubaluleke ngakumbi, i-AI sele ifumene izicelo ezibalulekileyo kukhuseleko lwesizwe, ukhuselo, kunye nemfazwe. Iziphumo zayo nayiphi na imfazwe yexesha elizayo ziya kuxhomekeka kakhulu ekwandiseni amandla okhuselo nge-AI; ngoko ke ii-arhente zikarhulumente ziya kuzama ukwakha amandla e-AI. Oku kwenza i-AI ibaluleke kakhulu kumazwe ekukhuselekeni kwesizwe.  

Iindlela ezikhoyo zokuqhubela phambili kwezobuchwepheshe zibonisa ukuba kungekudala kunokwenzeka ukuseka unxibelelwano olukhawulezayo phakathi kwengqondo kunye neeplatifomu ze-AI ezidlula i-interfaces zebhayoloji ezicothayo ukuze zidibanise ngempumelelo i-AI kwinqanaba lethu le-compute lesithathu. Amakhonkco e-neural ane-bandwidth ephezulu aya kusebenza njengebhulorho, edibanisa ngempumelelo ingqondo ne-AI. Abantu baya kuba zii-cyborgs (ii-cybernetic organisms). Ukudibana kuya kwenza ukuba ezi zimbini zizuze omnye komnye. Ingqondo iya kufumana amandla ekhompyutha angaphezu kwawemvelo e-AI ngaloo ndlela inciphise umngcipheko wokuba abantu baphelelwe lixesha xa bejongene nezidalwa zedijithali ezikrelekrele kakhulu. I-symbiosis yengqondo yomntu-AI iya kwenza abantu bakwazi ukulawula i-AI, ngaloo ndlela ibe yimpendulo kumngcipheko wobomi kuluntu obangelwa yi-AI ekrelekrele kakhulu.    

*** 

Imithombo:  

1. I-StarTalk (28 kaFebruwari 2026). Ngaba i-AI ifihla amandla ayo apheleleyo? NgoGeoffrey Hinton. Iyafumaneka apha https://www.youtube.com/watch?v=l6ZcFa8pybE 
2. Ulwazi lwaseKhanada ((27 Februwari 2026)). SIYITOAST: Utata we-AI uGeoffrey Hinton ulumkisa iSenate yaseKhanada ngengozi ekhoyo eluntwini. Ifumaneka apha https://www.youtube.com/watch?v=7fImPlfdRS0 
3. I-Neuralink. Uhlaziyo – Iminyaka emibini yeTelepathy. Iposiwe nge-28 kaJanuwari 2026. Iyafumaneka apha https://neuralink.com/updates/two-years-of-telepathy/ 
4. I-PRIME: Uphononongo Lokuqala Lokwenzeka Kwendlela Echanekileyo Yokunxibelelanisa Ubuchopho Nekhompyutha Efakwe NgeRobhothi Ukulawula Izixhobo Zangaphandle. Iyafumaneka apha  https://clinicaltrials.gov/study/NCT06429735
5. I-CONVOY: Uphononongo Lokuqala Lokwenzeka Kolawulo Lwee-Neural Zezixhobo Zokuncedisa Ngeteknoloji Yonxibelelwano Lwengqondo Nekhompyutha. Iyafumaneka apha https://clinicaltrials.gov/study/NCT06710626  
6. IZWI: Uphononongo Lokuqala Lokwenzeka Kwendlela Echanekileyo Yokudityaniswa Kwengqondo Nekhompyutha Efakwe NgeRobhothi Yokubuyisela Unxibelelwano. Iyafumaneka apha https://clinicaltrials.gov/study/NCT07224256 
7. ULex Fridman (2 Agasti 2024). U-Elon Musk: I-Neuralink kunye nekamva loLuntu. I-Podcast kaLex Fridman #438. Ifumaneka apha https://www.youtube.com/watch?v=Kbk9BiPhm7o 
8. UKumar, R., uWaisberg, E., uOng, J., kunye noLee, AG (2025). Amandla anokubakho eNeuralink - indlela unxibelelwano lobuchopho nomatshini olunokuguqula ngayo amayeza. Uhlolo lweeNgcali zeZixhobo zonyango, 22(6), 521–524. https://doi.org/10.1080/17434440.2025.2498457  
9. Bandre, P., et al 2025. “Neuralink: Ukuguqula iiBrain-Computer Interfaces zezeMpilo kunye nokuHlanganisa iHuman-AI,” 2025 INkomfa yesiBini yeHlabathi kwi-Electronic Circuits kunye neSignaling Technologies (ICECST), Petaling Jaya, Malaysia, 2025, iphe. 1122-1126, DOI: https://doi.org/10.1109/ICECST66106.2025.11307276 
10. I-UC Davis Health. Unxibelelwano olutsha lwekhompyutha nobuchopho luvumela indoda ene-ALS ukuba 'ithethe' kwakhona. 14 Agasti 2024. Iyafumaneka apha https://health.ucdavis.edu/news/headlines/new-brain-computer-interface-allows-man-with-als-to-speak-again/2024/08 
11. Vansteensel MJ, okqhubekayo 2016. Unxibelelwano lweNgqondo neKhompyutha olufakwe ngokupheleleyo kwisigulana esivalelwe ngaphakathi esine-ALS. N Engl J Med. 2016 Novemba 12;375(21):2060–2066. DOI: https://doi.org/10.1056/NEJMoa1608085 
12. UZhang X., okqhubekayo 2020. Indibaniselwano yonxibelelwano lwengqondo nekhompyutha kunye nobukrelekrele bokwenziwa: usetyenziso kunye nemingeni https://doi.org/10.21037/atm.2019.11.109 

*** 

Amanqaku afanelekileyo:  

Isifundo se-PRIME (Neuralink Clinical Trial): Umthathi-nxaxheba wesibini ufumana i-Implant (8 Agasti 2024)  

I-Neuralink: I-Neural Interface Elandelayo Enokutshintsha Ubomi Bomntu (29 Agasti 2020)  

I-BrainNet: Ityala lokuqala loNxibelelwano oluthe ngqo 'lweBrain-to-Brain' (5 Julayi 2019) 

***