Gartner Hype Cycle 2019: bayani

Mun warware fasahohin AI na 2019 kuma cikin rashin kunya mun kwatanta su da hasashen 2017.

Na farko, menene Gartner Hype Cycle? Wannan wani nau'i ne na zagayowar fasahar balagagge, ko kuma canji daga matakin zage-zage zuwa amfani mai amfani. Yanzu za a sami jadawali tare da fassarar don ƙara bayyanawa komai. Kuma a ƙasa akwai bayanin.

Matakin farko. fushi. Kaddamar. Fasahar ta bayyana, an fara tattauna ta ta hanyar masu wayewa, sannan kuma ta hanyar jama'a masu tsattsauran ra'ayi; a hankali farin ciki yana karuwa.

Mataki na biyu. ciniki. Kololuwar tsammanin kumbura. A wani lokaci, kowa ya riga ya yi magana game da fasaha, yana ƙoƙarin aiwatar da shi, kuma masu basira suna sayar da su a kan farashi mai tsada.

Mataki na uku. bakin ciki Rage sha'awa. Ana aiwatar da fasaha sosai kuma sau da yawa takan gaza saboda gazawa da iyakancewa. "Duk abin mamaki ne!" - zo nan da can. Abin farin ciki yana raguwa sosai (tambarin farashin, sau da yawa kuma).

Mataki na hudu. sabani Aiki a kan kwari. Ana inganta fasahar, ana magance matsalolin. A hankali, kamfanoni suna ƙoƙarin aiwatar da fasaha a hankali kuma, gaggawa, komai yana aiki sosai.

Mataki na biyar. Tallafawa Aiki mai albarka. Fasahar tana samun wurin da ta dace a kasuwa kuma tana aiki cikin nutsuwa, haɓakawa, kuma ana son ta.

Me ke faruwa?

Komawa zuwa zagayowar zagayowar 2019. Gartner saki a watan Satumba, wani rahoto kan abin da fasahar fasaha na wucin gadi ke a wane mataki, da kuma lokacin da za su fara aiki mai kyau. Hoton da ke ƙasa, sharhi a ƙasa jadawali.

Fasahar "Ganewar Magana" da "Haɓaka Tsari Ta Amfani da GPU" suna gaba da babban gefe kuma sun riga sun kasance a matakin "Ayyukan Samfura". Wannan yana nufin cewa dole ne a yi amfani da su cikin sauri, saboda sun riga sun ba da fa'ida ga masu mallakar su.

Koyon injina ta atomatik (AutoML) da chatbots a halin yanzu suna kan kololuwar hayaniya. Wato kowa yana magana game da su, da yawa suna aiwatar da su, amma zai ɗauki daga 2 zuwa 5 bisa sharadi don kawo fasahar zuwa daidaitattun da ake buƙata.

Motocin da muka saba yanzu ma sun fi na zamani. Fasahar abin hawa mai cin gashin kanta tana kusan gwada ƙasa. A wannan yanayin, wannan yana da kyau, saboda aiki mai amfani yana gaba. Koyaya, Gartner ya kiyasta cewa zai ɗauki akalla shekaru 10 don haɓakawa da daidaitawa.

Ina jirage marasa matuki da aka yi hasashe da gaske a yau? Komai yana cikin wurin - Gartner ya haɗa da jiragen sama marasa matuƙa a fagen Edge AI (rakunan da ke kan iyaka da AI), kuma gaskiyar kama-da-wane ta zama wani ɓangare na haɓakar hankali. Duk batutuwan biyu, ta hanyar, yanzu suna cikin matakin ƙaddamarwa kuma suna da tsinkaya mai kyau: 2-5 shekaru kafin aiki mai inganci akan kasuwa.

Abubuwan da suka dace

Daga cikin siffofi masu ban sha'awa: software na sarrafa kayan aikin robot - yana da ban tsoro, amma a gaskiya shi ne lokacin da mutum-mutumi ya maye gurbin ayyukan yau da kullum. Mafarki mai ban tsoro ga ƙananan ma'aikata; duk da haka binciken Harvard Business Review ya ce ba za a yi korafe-korafe ba, amma yawan aiki zai karu. Ku ci filaye yi imani. Fasahar za ta wuce kololuwar rashin amincewa da raini a cikin shekaru 2, sannan kuma ta yada ko'ina.

Daga cikin fasahohin da masu bishara da infogypsies na kowane ratsi za su yi magana game da masse kawai a nan gaba, "kayan aikin neuromorphic" ya kasance mai ban sha'awa musamman. Waɗannan na'urorin lantarki ne (chips) waɗanda koyi sifofin halittun halitta na tsarin jijiyarmu dangane da ingancin makamashi. Don sanya shi a sauƙaƙe, yana da game da babban aiki godiya ga rabon aiki (sabuntawa asynchronous na neurons). Kattai irin su IBM da Intel sun riga sun yi aiki tuƙuru wajen ƙirƙirar kwakwalwan kwamfuta na neuromorphic. Amma sojojin John Connor suna da lokacin shiryawa don ranar halaka - Gartner ya ba da fasahar har tsawon shekaru 10 don girma.

Yawanci, suna magana da yawa game da Da'a na Digital, amma ba sa gaggawar aiwatar da su. An keɓe jagorancin zuwa wani nau'i daban-daban na sassan AI: ana nufin cewa zai zama dole don ƙarfafa wasu ka'idoji na ɗabi'a, ka'idoji da ka'idoji don tattara bayanai, aiwatar da AI a cikin rayuwa, gabaɗaya, ta yadda zai kasance kamar haka. mutane. A ƙarshe, kalli Asimov.

2017 vs 2019

Yana da ban dariya, amma a cikin 2017 komai ya kasance daban, Babu ko da wani raba zagayowar hype ga AI: AI fasahar sun kasance a cikin locomotive na bunkasa fasahar (Emerging Technologies) tare da blockchain da ƙarin gaskiya.

Koyon na'ura da zurfin ilmantarwa sun kasance a kan ƙwaƙƙwaran Olympus a cikin 2017, kuma a cikin 2019 sun ci gaba da hanyarsu zuwa raguwa, wato. aiki mai albarka.

Af, jirage marasa matuka sun tashi daga kololuwa zuwa raguwa a cikin shekara, kuma a cikin 2019 sun koma kan gabatowa kololuwa. Kuma wannan yana faruwa, a.

A cikin 2019, zagayowar ya haɗa da sabbin fasahohi 8. Daga cikin su akwai sabis na girgije AI (Sabis na Cloud), Kasuwancin AI (Kasuwanci), Ƙididdigar Ƙididdigar da AI (Quantum Computing). Gabaɗaya, sanannun kayan aikin (a cikin kunkuntar da'ira) kayan aikin da ke fara sanya AI akan hanya.

source: www.habr.com