I-AI yezezimboni yintsimi ebanzi kune-intelligence equlethweyo, kwaye ubungakanani bayo bemarike bukhulu nangakumbi.

Iimeko zoshishino bezisoloko ziphakathi kwezona ndawo zibalulekileyo ekuthengisweni kwe-AI. Kwiminyaka emibini edlulileyo, iinkampani ezininzi ziqale ukusebenzisa kakhulu itekhnoloji ye-AI kwizixhobo, kumaziko edatha, nakwii-interfaces ze-human-machine ezikwi-intanethi (ii-HMI). Ngokwezibikezelo ze-IDC, izinga lokungena kunye nesabelo semarike ye-AI sikhula ngokukhawuleza, nokuba kwisoftware yenjongo ngokubanzi, isoftware yenkqubo yezoshishino, isoftware yombono wezimboni, okanye enye isoftware yeshishini. Ngokunjalo, imfuno yamandla ekhompyutha ye-AI, nokuba kwicala lesixhobo, kwicala lesixhobo, okanye kumaziko edatha, nayo iya kuqhubeka isanda.

Ukongeza kubukrelekrele obubonakalayo, iimeko ezithile zesicelo se-AI yemizi-mveliso ziquka oku kulandelayo:

Umbono womatshini: Njengovulindlela ekuphunyezweni kobukrelekrele bokwenziwa, iinethiwekhi ze-convolutional neural (ii-CNN) bezisetyenziswa kakhulu ekuhleleni nasekufumaneni kulo lonke uphuhliso lwazo. Kwiminyaka yakutshanje, ngokukhula kweemodeli ezinkulu, ubuchwepheshe be-AI obuninzi buye baziswa kumbono womatshini, njengokusebenzisa iimodeli zedatha enkulu ukuvelisa iisampulu okanye ukubona izinto ezingaqhelekanga, ngaloo ndlela boyisa imida yeenethiwekhi zemveli ze-CNN, ezifuna ukuqeqeshwa kwakhona xa zijongene neengxaki ezintsha.

Ulawulo lwemizi-mveliso: Kwicandelo lolawulo, ii-algorithms zakudala bezisetyenziswa ngaphambili. Nangona kunjalo, kwiminyaka emibini edlulileyo, ukufunda kokuqinisa kuye kwaba yindlela ethandwayo yokusasazwa, kokubini kulawulo lweerobhothi kunye nezinye izilawuli zemveli.

Ukwenziwa kwedijithali kwezoshishino: Ngaphambili, ingqwalasela yayikwimisebenzi efana nokuphucula ishedyuli yemveliso kunye nokusebenza kunye nokugcinwa kwayo. Le misebenzi, esekelwe ekufundeni koomatshini, iye yabandakanya kancinci ii-algorithms ze-AI ezingaphezulu. Kwiminyaka emibini edlulileyo, ukunyuka kweemodeli ezinkulu kukhokelele kwinkqubela phambili enkulu kwi-RAG (Retrieval Augmentation Generation). Abavelisi abaphambili kwihlabathi liphela, kunye nee-ODM ezininzi zasekhaya kunye nee-ISV, baye bamkela kakhulu i-RAG kwiimveliso zabo zesoftware ukunciphisa iindleko kunye nokwandisa ukusebenza kakuhle.