Rov qab mus rau qhov tseem ceeb, AIGC qhov kev ua tiav hauv kev sib piv yog kev sib txuas ntawm peb yam:

1. GPT yog ib tug replica ntawm tib neeg neurons

GPT AI sawv cev los ntawm NLP yog lub khoos phis tawj neural network algorithm, nws lub ntsiab lus yog simulate neural networks hauv tib neeg lub hlwb cortex.

Kev ua tiav thiab kev xav ntawm kev xav ntawm cov lus, suab paj nruag, duab, thiab cov ntaub ntawv txawm tias saj yog tag nrho cov haujlwm tsim los ntawm tib neeg.

lub hlwb raws li "protein computer" thaum lub sij hawm evolution mus ntev.

Yog li, GPT yog ib txwm tsim nyog tshaj plaws rau kev ua cov ntaub ntawv zoo sib xws, uas yog, cov lus tsis tsim nyog, suab paj nruag, thiab cov duab.

Lub tshuab ntawm nws cov txheej txheem tsis yog kev nkag siab ntawm lub ntsiab lus, tab sis yog txheej txheem ntawm kev ua kom zoo, txheeb xyuas, thiab koom nrog.Qhov no yog heev

paradoxical yam.

Early speech semantic recognition algorithms yeej tseem ceeb tsim qauv qauv sau ntawv thiab cov ntaub ntawv hais lus, tom qab ntawd kos cov lus rau cov lus,

tom qab ntawd muab cov lus tso rau hauv cov qauv sau ntawv kom nkag siab lub ntsiab lus ntawm cov lus, thiab thaum kawg tau txais kev lees paub.

Kev paub txog kev ua tau zoo ntawm qhov "kev xav txog lub tswv yim" raws li kev paub txog syntax tau hovering nyob ib ncig ntawm 70%, xws li ViaVoice lees paub

algorithm qhia los ntawm IBM hauv xyoo 1990s.

AIGC tsis yog hais txog kev ua si zoo li no.Nws lub ntsiab lus tsis yog xav txog cov qauv sau ntawv, tab sis yog tsim kom muaj ib lub neural network algorithm uas tso cai rau

computer suav cov kev sib txuas ntawm cov lus sib txawv, uas yog kev sib txuas ntawm neural, tsis yog kev sib txuas lus semantic.

Ntau yam zoo li kev kawm peb tus nplaig thaum peb tseem hluas, peb ib txwm kawm nws, tsis yog kawm "kev kawm, kwv yees, yam khoom, lus qhia, ntxiv,"

thiab ces nkag siab ib kab lus.

Qhov no yog tus qauv kev xav ntawm AI, uas yog kev lees paub, tsis nkag siab.

Qhov no kuj yog qhov tseem ceeb ntawm AI rau tag nrho cov qauv txheej txheem classical - cov khoos phis tawj tsis tas yuav nkag siab txog qhov teeb meem no ntawm qib kev xav,

tab sis theej txheeb xyuas thiab paub txog kev sib raug zoo ntawm cov ntaub ntawv sab hauv, thiab tom qab ntawd paub nws.

Piv txwv li, lub zog ntws ntawm lub xeev thiab kev twv ua ntej ntawm daim phiaj hluav taws xob yog ua raws li classical power network simulation, qhov twg tus qauv lej ntawm tus qauv.

mechanism yog tsim thiab ces converged siv ib tug matrix algorithm.Yav tom ntej, tej zaum nws yuav tsis tsim nyog.AI yuav ncaj qha txheeb xyuas thiab kwv yees a

qee tus qauv modal raws li cov xwm txheej ntawm txhua qhov ntawm.

Ntau cov nodes muaj, qhov tsawg nrov classical matrix algorithm yog, vim hais tias cov complexity ntawm lub algorithm nce nrog tus naj npawb ntawm cov algorithm.

nodes thiab geometric kev nce qib.Txawm li cas los xij, AI nyiam kom muaj qhov loj me ntawm qhov sib txuam, vim tias AI yog qhov zoo ntawm kev txheeb xyuas thiab

kwv yees qhov feem ntau yuav network hom.

Seb nws yog qhov kev kwv yees tom ntej ntawm Go (AlphaGO tuaj yeem kwv yees kaum ob ntawm cov kauj ruam tom ntej, nrog suav tsis txheeb rau txhua kauj ruam) lossis kev twv ua ntej modal

ntawm cov txheej txheem huab cua nyuaj, AI qhov tseeb yog ntau dua li cov qauv siv tshuab.

Yog vim li cas vim li cas daim phiaj hluav taws xob tam sim no tsis xav tau AI yog tias muaj pes tsawg leeg ntawm 220 kV thiab siab dua cov hluav taws xob sib txuas tswj hwm los ntawm lub xeev.

dispatching tsis loj, thiab ntau yam yog teem rau linearize thiab sparse lub matrix, zoo heev txo cov computational complexity.

mechanism qauv.

Txawm li cas los xij, nyob rau theem kev faib hluav taws xob hauv lub zog ntws, ntsib kaum txhiab lossis ntau pua txhiab lub hwj chim ntawm cov nodes, load nodes, thiab ib txwm muaj.

matrix algorithms nyob rau hauv ib tug loj tis network yog powerless.

Kuv ntseeg tias tus qauv lees paub ntawm AI ntawm qib kev faib khoom sib txuas yuav ua tau yav tom ntej.

2. Kev sib sau, kev cob qhia, thiab tsim cov ntaub ntawv tsis tsim nyog

Qhov thib ob yog vim li cas vim li cas AIGC tau ua qhov kev tawg yog kev khaws cov ntaub ntawv.Los ntawm A / D hloov dua siab tshiab ntawm kev hais lus (microphone + PCM

sampling) mus rau A / D hloov dua siab tshiab ntawm cov duab (CMOS + xim qhov chaw daim ntawv qhia), tib neeg tau sau cov ntaub ntawv holographic hauv qhov pom thiab hnov

teb nyob rau hauv txoj kev uas tsis tshua muaj nqi nyob rau hauv ob peb lub xyoo dhau los.

Hauv particular, qhov loj-loj nrov nrov ntawm cov koob yees duab thiab smartphones, cov ntaub ntawv khaws tseg ntawm cov ntaub ntawv tsis tsim nyog hauv thaj chaw audiovisual rau tib neeg.

ntawm yuav luag xoom tus nqi, thiab qhov tawg ntawm cov ntawv xov xwm hauv Is Taws Nem yog tus yuam sij rau AIGC kev cob qhia - cov ntaub ntawv kev cob qhia yog pheej yig.

Daim duab saum toj no qhia txog kev loj hlob ntawm cov ntaub ntawv thoob ntiaj teb, uas qhia meej txog kev nthuav dav.

Qhov kev loj hlob tsis-linear ntawm cov ntaub ntawv khaws cia yog lub hauv paus rau kev loj hlob tsis-linear ntawm AIGC lub peev xwm.

Tab sis, feem ntau ntawm cov ntaub ntawv no yog cov ntaub ntawv audio-visual unstructured, uas yog sau ntawm xoom nqi.

Hauv kev lag luam hluav taws xob, qhov no tsis tuaj yeem ua tiav.Ua ntej, feem ntau ntawm kev lag luam hluav taws xob hluav taws xob yog cov txheej txheem thiab cov ntaub ntawv semi-structured, xws li

voltage thiab tam sim no, uas yog cov ntsiab lus cov ntaub ntawv teev cov sij hawm series thiab semistructed.

Cov txheej txheem cov ntaub ntawv yuav tsum tau nkag siab los ntawm cov khoos phis tawj thiab xav tau "kev sib dhos", xws li kev sib dhos ntawm cov cuab yeej - cov ntaub ntawv voltage, tam sim no, thiab cov ntaub ntawv fais fab.

ntawm qhov hloov yuav tsum tau ua raws li qhov node.

Qhov teeb meem ntau dua yog lub sijhawm sib dhos, uas yuav tsum tau ua kom muaj zog, tam sim no, thiab lub zog ua haujlwm thiab rov ua haujlwm raws li lub sijhawm teev, yog li ntawd

Kev txheeb xyuas tom qab tuaj yeem ua tau.Kuj tseem muaj cov lus qhia rau pem hauv ntej thiab thim rov qab, uas yog qhov sib npaug ntawm qhov sib npaug ntawm plaub lub quadrant.

Tsis zoo li cov ntawv nyeem, uas tsis tas yuav tsum tau ua kom zoo, ib kab lus tsuas yog muab pov rau hauv lub computer, uas txheeb xyuas cov ntaub ntawv sib koom ua ke.

ntawm nws tus kheej.

Txhawm rau txhawm rau txhim kho qhov teeb meem no, xws li cov cuab yeej sib dhos ntawm cov ntaub ntawv kev faib khoom lag luam, kev sib raug zoo yog qhov xav tau tas li, vim tias qhov nruab nrab thiab

tsis tshua muaj hluav taws xob faib network yog ntxiv, tshem tawm, thiab hloov kho cov cuab yeej thiab kab txhua hnub, thiab cov tuam txhab kab sib chaws siv cov nqi ua haujlwm loj.

Zoo li "cov ntaub ntawv annotation," khoos phis tawj tsis tuaj yeem ua qhov no.

Qhov thib ob, tus nqi ntawm cov ntaub ntawv tau txais nyob rau hauv lub zog sector yog siab, thiab sensors yuav tsum tau es tsis txhob muaj lub xov tooj ntawm tes los hais lus thiab yees duab.”

Txhua lub sij hawm qhov voltage txo los ntawm ib theem (los yog lub hwj chim faib kev sib raug zoo txo ​​los ntawm ib theem), qhov yuav tsum tau sensor peev nce

los ntawm tsawg kawg ib qho kev txiav txim loj.Txhawm rau ua tiav qhov load sab (capillary kawg) kev paub, nws yog qhov ntau dua qhov kev nqis peev digital loj. "

Yog tias nws yog qhov tsim nyog los txheeb xyuas qhov hloov pauv hloov pauv ntawm daim phiaj hluav taws xob, qhov ntsuas qhov ntsuas qhov ntsuas siab yuav tsum tau ua, thiab tus nqi ntau dua.

Vim qhov tsis tshua muaj nqi marginal ntawm cov ntaub ntawv tau txais thiab cov ntaub ntawv sib dhos, daim phiaj hluav taws xob tam sim no tsis tuaj yeem ua kom txaus tsis-linear.

kev loj hlob ntawm cov ntaub ntawv cov ntaub ntawv los cob qhia ib qho algorithm kom ncav cuag AI singularity.

Tsis hais txog qhov qhib ntawm cov ntaub ntawv, nws tsis tuaj yeem ua rau lub zog AI pib kom tau txais cov ntaub ntawv no.

Yog li ntawd, ua ntej AI, nws yog ib qho tsim nyog los daws qhov teeb meem ntawm cov ntaub ntawv teev, txwv tsis pub AI code tsis tuaj yeem raug cob qhia los tsim AI zoo.

3. Kev tawg hauv kev suav lub zog

Ntxiv nrog rau cov algorithms thiab cov ntaub ntawv, qhov sib txawv ntawm qhov sib txawv ntawm AIGC kuj yog ib qho kev sib tw hauv kev suav lub zog.Ib txwm CPUs tsis yog

haum rau loj-scale concurrent neuronal xam.Nws yog qhov tseeb ntawm daim ntawv thov GPUs hauv 3D kev ua si thiab cov yeeb yaj kiab uas ua rau muaj qhov sib npaug loj

floating-point + streaming suav ua tau.Moore txoj cai ntxiv txo tus nqi suav rau ib chav tsev ntawm kev suav fais fab.

Daim phiaj hluav taws xob AI, qhov kev xav tsis tau yav tom ntej

Nrog rau kev sib koom ua ke ntawm ntau qhov kev faib tawm photovoltaic thiab faib lub zog cia, nrog rau cov ntawv thov uas yuav tsum tau ua.

load sab virtual fais fab nroj tsuag, nws yog objectively tsim nyog los ua qhov chaw thiab load forecasting rau pej xeem tis network systems thiab cov neeg siv.

kev faib (micro) kab sib chaws, nrog rau lub sijhawm tiag tiag lub zog txaus kom zoo rau kev faib (micro) kab sib chaws.

Kev sib txuas ntawm kev sib txuas ntawm kev faib tawm sab nraud yog qhov tseeb siab dua li ntawm kev teem caij sib kis.Txawm rau kev lag luam

complex, tej zaum yuav muaj kaum tawm txhiab ntawm cov khoom siv thauj khoom thiab ntau pua lub keyboards, thiab qhov kev thov rau AI raws li micro grid / faib network ua haujlwm

tswj yuav tshwm sim.

Nrog rau tus nqi qis ntawm cov sensors thiab kev siv dav siv hluav taws xob hluav taws xob xws li cov khoom siv hluav taws xob hauv lub xeev, cov khoom siv hluav taws xob hloov pauv, thiab cov khoom siv hluav taws xob (inverters),

kev sib koom ua ke ntawm kev paub, suav, thiab kev tswj hwm ntawm ntug ntawm daim phiaj hluav taws xob kuj tau dhau los ua ib qho kev hloov tshiab.

Yog li ntawd, AIGC ntawm daim phiaj hluav taws xob yog yav tom ntej.Txawm li cas los xij, qhov xav tau niaj hnub no tsis yog kom tshem tawm AI algorithm tam sim kom tau nyiaj,

Hloov chaw, ua ntej hais txog cov ntaub ntawv tsim kho cov teeb meem uas xav tau los ntawm AI

Hauv kev nce qib ntawm AIGC, yuav tsum muaj kev xav kom txaus siab txog qib kev thov thiab yav tom ntej ntawm lub zog AI.

Tam sim no, qhov tseem ceeb ntawm lub zog AI tsis yog qhov tseem ceeb: piv txwv li, photovoltaic algorithm nrog kev kwv yees qhov tseeb ntawm 90% yog muab tso rau hauv qhov chaw ua lag luam.

nrog rau kev sib txawv ntawm kev ua lag luam ntawm 5%, thiab qhov kev sib txawv ntawm algorithm yuav so tawm tag nrho cov txiaj ntsig kev lag luam.

Cov ntaub ntawv yog dej, thiab kev suav lub zog ntawm lub algorithm yog channel.Raws li nws tshwm sim, nws yuav yog.