Ke hoʻi nei i ke kumu, ʻo ka holomua o AIGC i ka singularity kahi hui o ʻekolu mau mea:

1. ʻO GPT kahi kope o nā neurons kanaka

ʻO GPT AI i hōʻike ʻia e NLP he algorithm network neural computer, nona ka mea nui e hoʻohālikelike i nā pūnaewele neural i loko o ka cerebral cortex kanaka.

ʻO ka hana a me ka noʻonoʻo naʻauao o ka ʻōlelo, nā mele, nā kiʻi, a me nā ʻike ʻono, nā hana āpau i hōʻiliʻili ʻia e ke kanaka.

lolo ma ke ʻano he "kamepiula protein" i ka wā lōʻihi.

No laila, ʻo GPT ka mea hoʻohālike kūpono loa no ka hoʻoponopono ʻana i ka ʻike like, ʻo ia hoʻi, ka ʻōlelo i kūkulu ʻole ʻia, nā mele, a me nā kiʻi.

ʻO ke ʻano o kāna hana ʻana ʻaʻole ia ka hoʻomaopopo ʻana i ke ʻano, akā he kaʻina hana no ka hoʻomaʻemaʻe, ʻike, a me ka hui ʻana.He mea nui loa kēia

mea kuʻikahi.

Ua ho'okumu mua 'ia nā algorithms e 'ike i ka semantic ha'i'ōlelo i ke kumu ho'ohālike grammar a me kahi waihona ha'i'ōlelo, a laila ho'opa'a 'ia ka ha'i'ōlelo i ka hua'ōlelo,

a laila hoʻokomo i ka huaʻōlelo i loko o ka waihona grammar no ka hoʻomaopopo ʻana i ke ʻano o ka huaʻōlelo, a loaʻa ka hopena ʻike.

ʻO ka maikaʻi o ka ʻike ʻana o kēia "logical mechanism" e pili ana i ka syntax e hoʻopaʻa nei i ka 70%, e like me ka ʻike ViaVoice.

algorithm i hoʻokomo ʻia e IBM i nā makahiki 1990.

ʻAʻole pili ʻo AIGC i ka pāʻani e like me kēia.ʻAʻole ka manaʻo nui o ka grammar, akā e hoʻokumu i kahi neural network algorithm e hiki ai i ka

kamepiula no ka helu ʻana i nā pilina pili probabilistic ma waena o nā huaʻōlelo like ʻole, ʻo ia nā pilina neural, ʻaʻole nā ​​pilina pili.

E like me ke aʻo ʻana i kā mākou ʻōlelo makuahine i ko mākou wā liʻiliʻi, ua aʻo maoli mākou iā ia, ma mua o ke aʻo ʻana i ka "kumu, predicate, object, verb, complement,"

a laila hoʻomaopopo i kahi paukū.

ʻO kēia ke kumu noʻonoʻo o AI, ʻo ia ka ʻike, ʻaʻole maopopo.

ʻO kēia hoʻi ke koʻikoʻi subversive o AI no nā hiʻohiʻona hana maʻamau - ʻaʻole pono nā kamepiula e hoʻomaopopo i kēia mea ma ka pae logical,

akā, e ʻike a ʻike i ka pilina ma waena o ka ʻike kūloko, a laila ʻike.

No ka laʻana, ua hoʻokumu ʻia ke kūlana kahe o ka mana a me ka wānana o nā pahu mana ma luna o ka hoʻohālikelike pūnaewele mana maʻamau, kahi kahi hiʻohiʻona makemakika o ka

Hoʻokumu ʻia ka mīkini a laila hoʻohui ʻia me ka hoʻohana ʻana i kahi algorithm matrix.I ka wā e hiki mai ana, ʻaʻole pono.E ʻike pono a wānana ʻo AI i a

kekahi kumu hoʻohālike e pili ana i ke kūlana o kēlā me kēia node.

ʻOi aku ka nui o nā nodes, ʻoi aku ka liʻiliʻi o ka matrix matrix algorithm, no ka mea, piʻi ka paʻakikī o ka algorithm me ka helu o

nodes a piʻi ka holomua geometric.Eia naʻe, makemake ʻo AI i ka loaʻa ʻana o ka node concurrency nui loa, no ka mea, maikaʻi ʻo AI i ka ʻike a me

e wānana ana i nā ʻano pūnaewele ʻoi loa.

Inā ʻo ia ka wānana aʻe o Go (hiki iā AlphaGO ke wānana i nā ʻanuʻu he nui e hiki mai ana, me nā mea hiki ʻole ke helu ʻia no kēlā me kēia ʻanuʻu) a i ʻole ka wānana modal

o nā ʻōnaehana ʻano paʻakikī, ʻoi aku ka kiʻekiʻe o ka pololei o AI ma mua o nā kumu hoʻohālike mechanical.

ʻO ke kumu i koi ʻole ʻia ai ka mana mana i kēia manawa AI ʻo ia ka helu o nā nodes ma 220 kV a ma luna o nā pūnaewele mana i mālama ʻia e ka panalāʻau.

ʻAʻole nui ka hoʻouna ʻana, a ua hoʻonohonoho ʻia nā kūlana he nui e linearize a hoʻokaʻawale i ka matrix, e hōʻemi nui ana i ka paʻakikī o ka helu ʻana.

kumu hoʻohālike mīkini.

Eia nō naʻe, ma ka pae hoʻoheheʻe ʻana o ka mana, e kū ana i nā ʻumi kaukani a i ʻole nā ​​haneli haneli o nā node mana, nā nodes hoʻouka, a me nā kuʻuna.

ʻAʻole mana ka matrix algorithms ma kahi pūnaewele hoʻolaha nui.

Ke manaʻoʻiʻo nei au e hiki ke ʻike ʻia ke kumu hoʻohālike o AI ma ka pae pūnaewele hoʻolaha i ka wā e hiki mai ana.

2. ʻO ka hōʻiliʻili ʻana, ka hoʻomaʻamaʻa ʻana, a me ka hoʻokumu ʻana o ka ʻike i kūkulu ʻole ʻia

ʻO ke kumu ʻelua i hana ai ʻo AIGC i kahi holomua ʻo ka hōʻiliʻili ʻana o ka ʻike.Mai ka hoʻololi ʻana o A/D o ka ʻōlelo (microphone+PCM

sampling) i ka hoʻololi A/D o nā kiʻi (CMOS+color space mapping), ua hōʻiliʻili nā kānaka i ka ʻikepili holographic i ka ʻike a me ka auditory.

nā māla i nā ala haʻahaʻa haʻahaʻa loa i nā makahiki i hala iho nei.

ʻO ka mea nui, ka hoʻolaha nui ʻana o nā kāmera a me nā smartphones, ka hōʻiliʻili ʻana o nā ʻikepili i hoʻonohonoho ʻole ʻia i ke kahua leo leo no nā kānaka.

ma kahi kokoke ʻole ke kumu kūʻai, a ʻo ka hōʻiliʻili pahū o ka ʻike kikokikona ma ka Pūnaewele ke kī i ka hoʻomaʻamaʻa AIGC - ʻaʻole maʻalahi nā hoʻonohonoho ʻikepili aʻo.

Hōʻike ka kiʻi ma luna nei i ka ulu ʻana o ka ʻikepili honua, e hōʻike maopopo ana i kahi ʻano exponential.

ʻO kēia ulu laina ʻole o ka hōʻiliʻili ʻikepili ke kumu no ka ulu ʻole laina o nā mana AIGC.

AKA, ʻo ka hapa nui o kēia mau ʻikepili he ʻikepili leo-kiʻi i kūkulu ʻole ʻia, i hōʻiliʻili ʻia ma ke kumu kūʻai ʻole.

Ma ke kahua o ka mana uila, ʻaʻole hiki ke hoʻokō ʻia kēia.ʻO ka mea mua, ʻo ka hapa nui o ka ʻoihana uila i hoʻonohonoho ʻia a me nā ʻikepili semi structured, e like me

Volta a me kēia manawa, ʻo ia nā pūʻulu ʻike kiko o ka manawa a me ka semi structured.

Pono e hoʻomaopopo ʻia nā pūʻulu ʻikepili i kūkulu ʻia e nā kamepiula a koi aku i ka "alignment", e like me ka alignment device - ka volta, ka manawa, a me ka ʻikepili mana.

pono e hoʻolikelike ʻia me kēia node.

ʻOi aku ka pilikia o ka hoʻoponopono ʻana i ka manawa, e koi ana i ka hoʻohālikelike ʻana i ka volta, ka manawa, a me ka mana hana a me ka reactive e pili ana i ka pae manawa, no laila

hiki ke hana i ka ʻike ma hope.Aia kekahi ʻaoʻao i mua a i hope, ʻo ia ka alignment spatial i ʻehā quadrants.

ʻAʻole like me ka ʻikepili kikokikona, ʻaʻole koi i ka alignment, hoʻolei wale ʻia kahi paukū i ke kamepiula, kahi e ʻike ai i nā hui ʻike hiki.

nona iho.

I mea e hoʻohālikelike ai i kēia pilikia, e like me ka alignment lako o ka ʻikepili hoʻohele ʻoihana, pono mau ka alignment, no ka mea, ʻo ka waena a me

Hoʻohui, holoi a hoʻololi i nā mea hana a me nā laina haʻahaʻa haʻahaʻa haʻahaʻa i kēlā me kēia lā, a hoʻolilo nā hui grid i nā koina hana nui.

E like me ka "annotation data," ʻaʻole hiki i nā kamepiula ke hana i kēia.

ʻO ka lua, he kiʻekiʻe ke kumukūʻai o ka loaʻa ʻana o ka ʻikepili ma ka ʻāpana mana, a koi ʻia nā mea ʻike ma mua o ka loaʻa ʻana o ke kelepona paʻa e kamaʻilio a paʻi kiʻi.”

I kēlā me kēia manawa e ho'ēmi ai ka voltage i hoʻokahi pae (a i ʻole ke emi ʻana o ka pilina hoʻoili mana i hoʻokahi pae), e hoʻonui ʻia ka hoʻokomo ʻana o ka sensor i makemake ʻia

ma ka liʻiliʻi hoʻokahi kauoha o ka nui.No ka hoʻokō ʻana i ka ʻaoʻao haʻahaʻa (capillary end) sensing, ʻoi aku ka nui o ka hoʻopukapuka kikohoʻe.

Inā pono e ʻike i ke ʻano transient o ka mana mana, koi ʻia ka sampling kiʻekiʻe-frequency kiʻekiʻe, a ʻoi aku ka kiʻekiʻe o ke kumukūʻai.

Ma muli o ke kumukūʻai marginal kiʻekiʻe loa o ka loaʻa ʻana o ka ʻikepili a me ka alignment data, ʻaʻole hiki i ka mana mana ke hōʻiliʻili i ka laina laina ʻole.

ka ulu ʻana o ka ʻikepili ʻikepili e hoʻomaʻamaʻa i kahi algorithm e hiki ai i ka AI singularity.

ʻAʻole e haʻi i ka wehe ʻana o ka ʻikepili, hiki ʻole i kahi mana AI hoʻomaka ke loaʻa i kēia mau ʻikepili.

No laila, ma mua o AI, pono e hoʻoponopono i ka pilikia o nā pūʻulu ʻikepili, inā ʻaʻole hiki ke hoʻomaʻamaʻa ʻia ke code AI maʻamau e hana i kahi AI maikaʻi.

3. Ka holomua ma ka mana helu

Ma waho aʻe o nā algorithms a me ka ʻikepili, ʻo ka holomua hoʻokahi o AIGC kahi holomua i ka mana helu.ʻAʻole nā ​​CPU kuʻuna

kūpono no ka nui-nui concurrent neuronal computing.ʻO ia ka hoʻohana pono ʻana o nā GPU i nā pāʻani 3D a me nā kiʻiʻoniʻoni e hoʻohālikelike ai i ka nui.

floating-point+streaming computing hiki.Hoʻemi hou ke kānāwai o Moore i ke kumukūʻai helu no kēlā me kēia ʻāpana o ka mana helu.

Power grid AI, kahi ʻano hiki ʻole i ka wā e hiki mai ana

Me ka hoʻohui ʻana o ka nui o nā photovoltaic i hoʻolaha ʻia a me nā ʻōnaehana mālama ikehu, a me nā koi noi o

ʻO ka hoʻouka ʻana i nā mea kanu mana virtual, pono ia e alakaʻi i ka wanana kumu a me ka hoʻouka ʻana no nā ʻōnaehana hoʻolaha lehulehu a me ka mea hoʻohana

nā ʻōnaehana mākia hoʻohelehele (micro), a me ka hoʻomaʻamaʻa ʻana i ka mana kahe o ka manawa maoli no ka hāʻawi ʻana i nā ʻōnaehana mānoanoa (micro).

ʻOi aku ka nui o ka paʻakikī helu o ka ʻaoʻao pūnaewele hoʻolaha ma mua o ka hoʻonohonoho ʻana o ka pūnaewele hoʻouna.ʻOiai no kahi kālepa

paʻakikī, aia paha he ʻumi kaukani o nā hāmeʻa hoʻouka a me nā haneli o nā hoʻololi, a me ke koi no ka hana ʻana o ka micro grid/distribution network e pili ana i AI.

e ala mai ka mana.

Me ke kumu kūʻai haʻahaʻa o nā mea ʻike a me ka hoʻohana nui ʻana i nā mea uila uila e like me nā transformers solid-state, nā hoʻololi paʻa paʻa, a me nā mea hoʻohuli (converters),

ʻo ka hoʻohui ʻana o ka sensing, computing, a me ka hoʻomalu ʻana ma ka ʻaoʻao o ka mana mana i lilo i mea hou.

No laila, ʻo ka AIGC o ka mana mana ka wā e hiki mai ana.Eia naʻe, ʻo ka mea e pono ai i kēia lā ʻaʻole e lawe koke i kahi algorithm AI e loaʻa kālā,

Ma kahi, e kamaʻilio mua i nā pilikia kūkulu ʻikepili i koi ʻia e AI

I ka piʻi ʻana o AIGC, pono e lawa ka noʻonoʻo mālie e pili ana i ka pae noi a me ka wā e hiki mai ana o ka mana AI.

I kēia manawa, ʻaʻole koʻikoʻi ka manaʻo o ka mana AI: no ka laʻana, kahi photovoltaic algorithm me ka wānana pololei o 90% i kau ʻia ma ka mākeke wahi.

me ka paepae hoʻokaʻawale kālepa o 5%, a ʻo ka algorithm deviation e holoi i nā waiwai kālepa āpau.

ʻO ka ʻikepili ka wai, a ʻo ka mana helu o ka algorithm he kahawai.E like me ka hana ʻana, e hana ʻia.