E pili ana i ka Bias Artificial Intelligence

ll:

• ʻIke ʻia ke aʻo ʻana i nā mīkini i nā mamana i ka ʻikepili. Akā, hiki ke "biased" -ʻo ia hoʻi, e huli i nā mamana i pololei ʻole. No ka laʻana, hiki i kahi ʻōnaehana ʻike maʻi maʻi ʻili ma ke kiʻi ke nānā kūikawā i nā kiʻi i lawe ʻia ma ke keʻena o ke kauka. ʻAʻole hiki ke aʻo mīkini e maopopo ai: ʻike wale kāna mau algorithms i nā mamana i nā helu, a inā ʻaʻole hōʻike ka ʻikepili, pēlā nō ka hopena o kāna hana. A paʻakikī paha ka hopu ʻana i ia mau pōpoki ma muli o ke ʻano mechanics o ke aʻo ʻana i ka mīkini.
• ʻO ka ʻaoʻao pilikia a paʻakikī loa ka ʻokoʻa kanaka. Nui nā kumu e nalowale ai ka ʻikepili e pili ana i nā kānaka ma ka pae o ka hōʻiliʻili ʻana. Mai noʻonoʻo e pili wale kēia pilikia i ka poʻe: e kū like nā pilikia like i ka wā e hoʻāʻo ai e ʻike i kahi kahe i loko o kahi hale kūʻai a i ʻole ka turbine kinoea. Hiki ke hoʻohālikelike ʻia kekahi mau ʻōnaehana i ka waihoʻoluʻu o ka ʻili, ʻo nā mea ʻē aʻe e pili ana i nā mea ʻike Siemens.
• ʻAʻole ia mau pilikia hou i ka aʻo ʻana i ka mīkini, a mamao loa lākou mai kahi kū hoʻokahi. Hana ʻia nā kuhi hewa i loko o kekahi ʻano paʻakikī, a paʻakikī mau ka hoʻomaopopo ʻana i ke kumu i hoʻoholo ʻia ai. Pono mākou e hakakā i kēia ma ke ʻano holoʻokoʻa: hana i nā mea hana a me nā kaʻina no ka hōʻoia - a aʻo i nā mea hoʻohana i ʻole lākou e hahai makapō i nā ʻōlelo aʻoaʻo AI. ʻOi aku ka maikaʻi o ke aʻo ʻana i nā mīkini ma mua o ka hiki iā mākou - akā, ʻoi aku ka maikaʻi o nā ʻīlio ma mua o ke kanaka i ka ʻike ʻana i nā lāʻau lapaʻau, ʻaʻole ia he kumu e hoʻohana ai iā lākou i mea hōʻike a hoʻoholo i kā lākou hōʻike. A ʻo nā ʻīlio, ma ke ala, ʻoi aku ka naʻauao ma mua o nā ʻōnaehana aʻo mīkini.

ʻO ke aʻo ʻana i ka mīkini kekahi o nā ʻano ʻenehana kumu nui i kēia lā. ʻO kēia kekahi o nā ala nui e hoʻololi ai ka ʻenehana i ka honua a puni mākou i nā makahiki he ʻumi e hiki mai ana. ʻO kekahi mau hiʻohiʻona o kēia mau hoʻololi ke kumu e hopohopo ai. No ka laʻana, ka hopena kūpono o ke aʻo ʻana o ka mīkini ma ka mākeke hana, a i ʻole kona hoʻohana ʻana no nā kumu pono ʻole (e laʻana, e nā aupuni mana). Aia kekahi pilikia ʻē aʻe e kamaʻilio nei kēia pou: hoʻokaʻawale naʻauao.

ʻAʻole kēia he moʻolelo maʻalahi.

Hiki i ka AI o Google ke loaʻa nā pōpoki. He mea kūikawā kēia nūhou mai ka makahiki 2012 mai ia manawa.

He aha ka "AI Bias"?

ʻO "Raw Data" he oxymoron a he manaʻo maikaʻi ʻole; pono e hoʻomākaukau maikaʻi ʻia nā ʻikepili. —Geoffrey Bocker

Ma kahi ma mua o 2013, i mea e hana ai i kahi ʻōnaehana, e ʻōlelo, e ʻike i nā pōpoki i nā kiʻi, pono ʻoe e wehewehe i nā pae kūpono. Pehea e ʻike ai i nā kihi o ke kiʻi, e ʻike i nā maka, e kālailai i nā ʻano huluhulu, e helu i nā wāwae, a pēlā aku. A laila e hoʻohui i nā ʻāpana āpau a ʻike ʻaʻole pono ia. E like me ka lio mechanical - hiki ke hana ʻia, akā ma ka hoʻomaʻamaʻa he paʻakikī ke wehewehe. ʻO ka hopena he mau haneli (a he mau kaukani paha) o nā lula kākau lima. A ʻaʻole hoʻokahi kumu hoʻohālike hana.

Me ka hiki ʻana mai o ke aʻo ʻana o ka mīkini, ua hoʻōki mākou i ka hoʻohana ʻana i nā lula "manual" no ka ʻike ʻana i kahi mea. Akā, lawe mākou i hoʻokahi kaukani laʻana o "kēia", X, hoʻokahi tausani laʻana o "nā mea ʻē aʻe", Y, a hana ke kamepiula i kumu hoʻohālike e pili ana i kā lākou ʻikepili helu. A laila hāʻawi mākou i kēia ʻano hoʻohālike i kekahi mau laʻana ʻikepili a hoʻoholo me ka pololei inā kūpono ia i kekahi o nā pūʻulu. Hoʻokumu ke aʻo mīkini i kahi kumu hoʻohālike mai ka ʻikepili ma mua o ke kākau ʻana a ke kanaka. He mea kupaianaha nā hopena, ʻoi aku hoʻi ma ke kahua o ke kiʻi a me ka ʻike kumu, a ʻo ia ke kumu e neʻe nei ka ʻoihana ʻenehana holoʻokoʻa i ke aʻo mīkini (ML).

Akā ʻaʻole ia maʻalahi. Ma ke ao maoli, ʻo kāu mau kaukani laʻana o X a i ʻole Y kekahi i loaʻa iā A, B, J, L, O, R, a me L. ʻAʻole hiki ke puunaue like ʻia kēia mau mea, a hiki ke hana pinepine ʻia kekahi e uku nui ai ka ʻōnaehana. e nānā iā lākou ma mua o nā mea hoihoi iā ʻoe.

He aha ke ʻano o kēia ma ka hana? ʻO kaʻu hiʻohiʻona punahele ʻo ia ka wā ʻōnaehana ʻike kiʻi e nānā i kahi puʻu mauʻu a ʻōlelo, "hipa". Ua maopopo ke kumu: lawe ʻia ka hapa nui o nā kiʻi laʻana o nā "hipa" ma nā kula kahi a lākou e noho ai, a ma kēia mau kiʻi e ʻoi aku ka nui o ka mauʻu ma mua o nā hulu keʻokeʻo liʻiliʻi, a ʻo ka mauʻu ka mea i manaʻo nui ʻia e ka ʻōnaehana. .

Aia kekahi mau hiʻohiʻona koʻikoʻi. Hoʻokahi mea hou o ka hana no ka ʻike ʻana i ka maʻi kanesa ʻili ma nā kiʻi. Ua ʻike ʻia ua kiʻi pinepine nā dermatologists i ke aliʻi me nā hōʻike o ka maʻi maʻi ʻili e hoʻopaʻa i ka nui o nā ʻano. ʻAʻohe aliʻi i nā kiʻi laʻana o ka ʻili olakino. No kahi ʻōnaehana AI, ua lilo ia mau aliʻi (ʻoi aku ka pololei, nā pika a mākou e wehewehe ai he "ruler") i lilo i hoʻokahi o nā ʻokoʻa ma waena o nā hoʻohālike, a i kekahi manawa ʻoi aku ka nui ma mua o kahi ʻōpala liʻiliʻi ma ka ʻili. No laila, ua hana ʻia kahi ʻōnaehana e ʻike ai i ka maʻi maʻi ʻili i kekahi manawa.

ʻO ka mea nui ma ʻaneʻi, ʻaʻohe ʻike o ka ʻōnaehana i kāna mea e nānā nei. Nānā mākou i kahi pūʻulu pixels a ʻike i loko o lākou he hipa, ʻili a i ʻole nā ​​aliʻi, akā ʻo ka ʻōnaehana he laina helu wale nō. ʻAʻole ʻike ʻo ia i ke ākea ʻekolu, ʻaʻole ʻike i nā mea, nā ʻano, a me nā hipa. ʻIke wale ʻo ia i nā mamana i ka ʻikepili.

ʻO ka paʻakikī o ka ʻike ʻana i ia mau pilikia ʻo ia ka neural network (ke kumu hoʻohālike i hana ʻia e kāu ʻōnaehana aʻo mīkini) he mau kaukani o nā haneli haneli. ʻAʻohe ala maʻalahi e nānā i kahi hoʻohālike a ʻike pehea e hoʻoholo ai. ʻO ka loaʻa ʻana o ia ʻano he mea maʻalahi ke kaʻina hana e wehewehe i nā lula āpau me ka lima, me ka ʻole o ka hoʻohana ʻana i ka mīkini aʻo. Ke hopohopo nei ka poʻe ua lilo ke aʻo ʻana i ka mīkini i mea o ka pahu ʻeleʻele. (E wehewehe liʻiliʻi wau ma hope no ke aha i ʻoi loa ai kēia hoʻohālikelike.)

ʻO kēia, ma nā ʻōlelo maʻamau, ʻo ia ka pilikia o ka hoʻokaʻawale ʻana i ka naʻauao a i ʻole ke aʻo ʻana i ka mīkini: ʻike paha kahi ʻōnaehana no ka ʻimi ʻana i nā mamana i ka ʻikepili i nā hiʻohiʻona hewa, a ʻaʻole paha ʻoe e ʻike. He hiʻohiʻona koʻikoʻi kēia o ka ʻenehana, a ʻike ʻia i nā poʻe a pau e hana pū me ia ma ke kula a me nā ʻoihana ʻenehana nui. Akā paʻakikī kona hopena, a pēlā nō kā mākou hoʻonā hiki i kēlā mau hopena.

E kamaʻilio mua kākou i nā hopena.

Hiki iā AI, implicitly no mākou, ke koho i kahi koho no kekahi mau ʻano o ka poʻe, e pili ana i ka nui o nā hōʻailona ʻike ʻole.

AI Bias Scenarios

ʻO ka mea maopopo loa a me ka weliweli, hiki i kēia pilikia ke hōʻike iā ia iho i ka wā e pili ana i ka ʻokoʻa kanaka. I kēia mau lā he lonoua hoʻāʻo ʻo Amazon e kūkulu i kahi ʻōnaehana aʻo mīkini no ka nānā mua ʻana i nā moho hana. Ma muli o ka nui o nā kāne ma waena o nā limahana Amazon, ʻo nā hiʻohiʻona o ka "hoʻolimalima kūleʻa" ʻoi aku ka nui o nā kāne, a ʻoi aku ka nui o nā kāne i ke koho ʻana i nā hoʻomaka i manaʻo ʻia e ka ʻōnaehana. Ua ʻike ʻo Amazon i kēia a ʻaʻole i hoʻokuʻu i ka ʻōnaehana i ka hana.

ʻO ka mea nui loa i kēia hiʻohiʻona ʻo ia ka ʻōlelo ʻana o ka ʻōnaehana e makemake i nā mea noi kāne, ʻoiai ʻaʻole i kuhikuhi ʻia ke kāne ma ka resume. Ua ʻike ka ʻōnaehana i nā hiʻohiʻona ʻē aʻe i nā laʻana o "nā hoʻolimalima maikaʻi": no ka laʻana, hiki i nā wahine ke hoʻohana i nā huaʻōlelo kūikawā e wehewehe i nā hana, a i ʻole nā ​​​​leʻaleʻa kūikawā. ʻOiaʻiʻo, ʻaʻole ʻike ka ʻōnaehana i ke ʻano o ka "hockey", a ʻo wai ka "poʻe", a i ʻole ka "holomua" - ua hoʻokō wale ʻia kahi ʻikepili helu o ka kikokikona. Akā, ʻaʻole ʻike ʻia e ke kanaka nā hiʻohiʻona āna i ʻike ai, a ʻo kekahi o lākou (e like me ka wehewehe ʻana o ka poʻe o nā kāne like ʻole i ka holomua) he paʻakikī paha iā mākou ke ʻike inā mākou e nānā iā lākou.

ʻOi aku - ʻoi aku ka ʻino. ʻO kahi ʻōnaehana aʻo mīkini maikaʻi loa i ka loaʻa ʻana o ka maʻi maʻi ma ka ʻili haikea ʻaʻole hiki ke hana maikaʻi ma ka ʻili ʻeleʻele, a i ʻole. ʻAʻole pono ma muli o ke ʻano, akā no ka mea pono paha ʻoe e kūkulu i kahi hiʻohiʻona kaʻawale no kahi kala ʻili ʻē aʻe, e koho ana i nā ʻano like ʻole. ʻAʻole hiki ke hoʻololi ʻia nā ʻōnaehana aʻo mīkini ʻoiai ma kahi wahi haiki e like me ka ʻike kiʻi. Pono ʻoe e hoʻololi i ka ʻōnaehana, i kekahi manawa ma o ka hoʻāʻo ʻana a me ka hewa, e kiʻi i kahi kikoʻī maikaʻi i nā hiʻohiʻona i ka ʻikepili āu e makemake ai a hiki i ka loaʻa ʻana o ka pololei āu e makemake ai. Akā ʻo ka mea āu e ʻike ʻole ai he pololei ka ʻōnaehana 98% o ka manawa me hoʻokahi hui, a ʻo 91% wale nō (ʻoi aku ka pololei ma mua o ka loiloi kanaka) me kekahi.

I kēia manawa ua hoʻohana nui au i nā hiʻohiʻona e pili ana i nā kānaka a me ko lākou ʻano. ʻO ke kūkākūkā e pili ana i kēia pilikia e pili nui ana i kēia kumuhana. Akā, he mea nui ka hoʻomaopopo ʻana he ʻāpana wale nō ka manaʻo o ka poʻe i ka pilikia. E hoʻohana mākou i ke aʻo ʻana i ka mīkini no nā mea he nui, a pili ka sampling error iā lākou āpau. Ma ka ʻaoʻao ʻē aʻe, inā ʻoe e hana pū me nā poʻe, ʻaʻole pili paha ka manaʻo o ka ʻikepili iā lākou.

No ka hoʻomaopopo ʻana i kēia, e hoʻi kākou i ka laʻana o ka maʻi maʻi ʻili a noʻonoʻo i ʻekolu mau manaʻo hypothetical no ka hemahema o ka ʻōnaehana.

1. Heterogeneous mahele ʻana o nā kānaka: he helu kaulike ʻole o nā kiʻi o nā ʻano ʻili like ʻole, e alakaʻi ana i nā hopena maikaʻi ʻole a i ʻole nā ​​​​hewa hewa ʻole ma muli o ka pigmentation.
2. Aia i loko o ka ʻikepili kahi i aʻo ʻia ai ka ʻōnaehana kahi hiʻohiʻona i hoʻolaha pinepine ʻia a heterogeneously ʻaʻole pili me nā kānaka a ʻaʻohe waiwai diagnostic: he aliʻi i nā kiʻi o ka maʻi maʻi ʻili a i ʻole ka mauʻu i nā kiʻi hipa. I kēia hihia, e ʻokoʻa ka hopena inā ʻike ka ʻōnaehana i nā pixels i ke kiʻi o kahi mea i ʻike ʻia e ka maka kanaka he "ruler".
3. Aia ka ʻikepili i kahi ʻano ʻaoʻao ʻekolu ʻaʻole hiki i ke kanaka ke ʻike inā ʻimi ʻo ia.

He aha ka manaʻo? Ua ʻike mua mākou e hōʻike ana ka ʻikepili i nā hui like ʻole o ka poʻe, a ma ka liʻiliʻi hiki iā mākou ke hoʻolālā e ʻimi i ia mau ʻokoʻa. I nā huaʻōlelo ʻē aʻe, nui nā kumu pilikanaka e manaʻo ai ua loaʻa i ka ʻikepili e pili ana i nā pūʻulu o ka poʻe i kekahi mau manaʻo. Inā mākou e nānā i ke kiʻi me ke aliʻi, e ʻike mākou i kēia aliʻi - ua nānā wale mākou iā ia ma mua, me ka ʻike ʻole he mea nui ia, a poina ʻaʻole ʻike ka ʻōnaehana i kekahi mea.

Akā, pehea inā i lawe ʻia kāu mau kiʻi āpau o ka ʻili maikaʻi ʻole i loko o kahi keʻena ma lalo o ke kukui uila, a ua lawe ʻia kou ʻili olakino ma lalo o ke kukui fluorescent? He aha inā, ma hope o kou pana ʻana i ka ʻili olakino, ma mua o ka pana ʻana i ka ʻili maikaʻi ʻole, ua hōʻano hou ʻoe i ka ʻōnaehana hana ma kāu kelepona, a ua hoʻololi iki ʻo Apple a i ʻole Google i ka algorithm noise reduction? ʻAʻole hiki i ke kanaka ke hoʻomaopopo i kēia, no ka nui o kāna ʻimi ʻana i ia mau hiʻohiʻona. Akā e ʻike koke ka ʻōnaehana hoʻohana mīkini a hoʻohana i kēia. ʻAʻole ʻo ia i ʻike i kekahi mea.

I kēia manawa ua kamaʻilio mākou e pili ana i nā hoʻoponopono hoʻopunipuni, akā pololei paha ka ʻikepili a pololei nā hopena, akā ʻaʻole ʻoe makemake e hoʻohana iā lākou no nā kumu etika, kānāwai, a i ʻole hoʻokele. ʻAʻole ʻae ʻia kekahi mau ʻāina, no ka laʻana, ʻaʻole i ʻae nā wahine e loaʻa ka uku hoʻemi ma kā lākou ʻinikua, ʻoiai ʻoi aku ka palekana o nā wahine. Hiki iā mākou ke noʻonoʻo i kahi ʻōnaehana, i ka wā e nānā ana i ka ʻikepili mōʻaukala, e hāʻawi i kahi kumu pilikia haʻahaʻa i nā inoa wahine. ʻAe, wehe mākou i nā inoa mai ka koho. Akā, e hoʻomanaʻo i ka laʻana Amazon: hiki i ka ʻōnaehana ke hoʻoholo i ke kāne ma muli o nā kumu ʻē aʻe (ʻoiai ʻaʻole ʻo ia i ʻike i ke ʻano o ke kāne, a i ʻole ke ʻano o ke kaʻa), a ʻaʻole ʻoe e ʻike i kēia a hiki i ka hoʻoponopono hou ʻana i nā kumukūʻai āu. hāʻawi a hoʻopaʻi iā ʻoe e hoʻopaʻi ʻia ʻoe.

ʻO ka hope, manaʻo pinepine ʻia e hoʻohana wale mākou i ia mau ʻōnaehana no nā papahana e pili ana i nā poʻe a me nā pilina pili. He hewa kēia. Inā hana ʻoe i nā turbines kinoea, makemake paha ʻoe e hoʻopili i ke aʻo ʻana i ka mīkini i ka telemetry i hoʻouna ʻia e nā ʻumi a i ʻole haneli mau mea ʻike ma kāu huahana (leo, wikiō, wela, a me nā mea ʻike ʻē aʻe e hoʻopuka i ka ʻikepili hiki ke maʻalahi ke hoʻololi ʻia e hana i kahi mīkini. kumu hoʻohālike ). Hiki iā ʻoe ke ʻōlelo, "Eia ka ʻikepili mai hoʻokahi kaukani turbine i hāʻule ma mua o ka hāʻule ʻana, a eia ka ʻikepili mai hoʻokahi tausani turbines i hāʻule ʻole. E kūkulu i kahi hoʻohālike e haʻi i ka ʻokoʻa ma waena o lākou. " ʻĀ, e noʻonoʻo ʻoe ua kau ʻia nā sensor Siemens ma 75% o nā turbines maikaʻi ʻole, a ʻo 12% wale nō o nā mea maikaʻi (ʻaʻohe pili me nā hemahema). E kūkulu ka ʻōnaehana i kumu hoʻohālike e ʻimi ai i nā turbines me nā mea ʻike Siemens. Auwe!

Kiʻi — Moritz Hardt, UC Berkeley

Ka mālama ʻana iā AI Bias

He aha kā mākou e hana ai no ia mea? Hiki iā ʻoe ke hoʻokokoke i ka pilikia mai ʻekolu ʻaoʻao:

1. ʻO ke koʻikoʻi me ka ʻohi ʻana a me ka mālama ʻana i ka ʻikepili no ke aʻo ʻana i ka ʻōnaehana.
2. Nā mea hana ʻenehana no ka nānā ʻana a me ka ʻike ʻana i ke ʻano hoʻohālike.
3. E hoʻomaʻamaʻa, hoʻonaʻauao, a e akahele i ka hoʻokō ʻana i ke aʻo ʻana i ka mīkini i nā huahana.

He ʻakaʻaka ma ka puke a Molière "The Bourgeois in the Nobility": ua haʻi ʻia kekahi kanaka ua māhele ʻia ka palapala i ka prose a me nā mele, a ua hauʻoli ʻo ia i ka ʻike ʻana ua kamaʻilio ʻo ia ma ka moʻolelo i kona ola a pau, me ka ʻike ʻole. ʻO kēia paha ka manaʻo o ka poʻe helu helu i kēia lā: me ka ʻike ʻole, ua hoʻolaʻa lākou i kā lākou ʻoihana i ka naʻauao akamai a me ka hoʻohālikelike ʻana i ka hewa. ʻO ka ʻimi ʻana i ka hewa o ka laʻana a me ka hopohopo e pili ana iā ia ʻaʻole ia he pilikia hou, pono mākou e hoʻokokoke systematically i kāna hopena. E like me ka mea i ʻōlelo ʻia ma luna, i kekahi mau mea ʻoi aku ka maʻalahi o ka hana ʻana i kēia ma ke aʻo ʻana i nā pilikia pili i ka ʻikepili kanaka. Manaʻo mākou he manaʻo ʻino paha mākou e pili ana i nā hui like ʻole o ka poʻe, akā paʻakikī iā mākou ke noʻonoʻo i kahi manaʻo ʻino e pili ana i nā sensor Siemens.

ʻO ka mea hou e pili ana i kēia mau mea, ʻoiaʻiʻo, ʻaʻole hana pololei nā kānaka i ka ʻikepili helu. Lawe ʻia ia e nā mīkini e hana i nā hiʻohiʻona nui a paʻakikī i paʻakikī ke hoʻomaopopo. ʻO ka hoʻopuka ʻana i ka ʻike maopopo ʻo ia kekahi o nā mea nui o ka pilikia o ka bias. Ke hopohopo nei mākou ʻaʻole pili wale ka ʻōnaehana, akā ʻaʻohe ala e ʻike ai i kona ʻano, a ʻokoʻa ke aʻo ʻana o ka mīkini mai nā ʻano ʻē aʻe o ka automation, i manaʻo ʻia he mau ʻanuʻu kūpono e hiki ke hoʻāʻo ʻia.

Elua pilikia maanei. Hiki iā mākou ke hana i kekahi ʻano loiloi o nā ʻōnaehana aʻo mīkini. A ʻaʻole maʻalahi ka loiloi ʻana i nā ʻōnaehana ʻē aʻe.

ʻO ka mea mua, ʻo kekahi o nā kuhikuhi o ka noiʻi hou i ke kula o ka mīkini aʻo ʻo ia ka ʻimi ʻana i nā ala e ʻike ai i ka hana nui o nā ʻōnaehana aʻo mīkini. ʻO kēlā ʻōlelo, ʻo ke aʻo ʻana i ka mīkini (ma kona kūlana i kēia manawa) he ʻano ʻepekema hou loa e loli wikiwiki nei, no laila, mai manaʻo ʻoe ʻaʻole hiki ke lilo koke nā mea hiki ʻole i kēia lā. Papahana Aia - he laʻana hoihoi o kēia.

ʻO ka lua, ʻo ka manaʻo e hiki iā ʻoe ke hoʻāʻo a hoʻomaopopo i ke kaʻina hana hoʻoholo o nā ʻōnaehana a i ʻole nā ​​hui e kū nei he maikaʻi i ke kumumanaʻo, akā ma ka hana. ʻAʻole maʻalahi ka hoʻomaopopo ʻana i ka hoʻoholo ʻana i loko o kahi hui nui. ʻOiai inā he kaʻina hana hoʻoholo maʻamau, ʻaʻole ia e hōʻike i ke ʻano o ka launa ʻana o ka poʻe, a ʻaʻole pinepine lākou i ke ala kūpono a me ka ʻōnaehana e hoʻoholo ai. E like me kā koʻu hoa hana i ʻōlelo ai ʻO Vijay Pande, He pahu ʻeleʻele hoʻi nā kānaka.

E lawe i hoʻokahi kaukani poʻe i loko o kekahi mau hui a me nā ʻoihana, a ʻoi aku ka paʻakikī o ka pilikia. Ua ʻike mākou ma hope o ka hoʻomaka ʻana o ka Space Shuttle e wāwahi i ka hoʻi ʻana mai, a loaʻa i nā poʻe i loko o NASA ka ʻike i hāʻawi iā lākou i kumu e noʻonoʻo ai e hiki mai ana kekahi mea ʻino, akā ʻo ka ʻōnaehana. laulā ʻAʻole au i ʻike i kēia. Ua hele wale ʻo NASA i kahi loiloi like ma hope o ka nalowale ʻana o kāna kaʻa kaʻa mua, akā naʻe ua nalowale kekahi no ke kumu like. He mea maʻalahi ka hoʻopaʻapaʻa ʻana e hahai nā hui a me nā kānaka i nā lula kūpono a hiki ke hoʻāʻo ʻia, hoʻomaopopo ʻia a hoʻololi ʻia—akā, ʻokoʻa ka ʻike. ʻO kēia "ʻO ka hoʻopunipuni a Gosplan".

Hoʻohālikelike pinepine au i ka aʻo ʻana i nā mīkini me nā ʻikepili, ʻoi aku ka pili - he ʻenehana kumu hou i hoʻololi i nā hiki o ka ʻepekema kamepiula a me ka honua a puni, i lilo i ʻāpana o nā mea āpau, a mākou e hoʻohana mau ai me ka ʻole o ka ʻike. Loaʻa i nā ʻikepili nā pilikia, a he ʻano like lākou: hiki ke kūkulu ʻia ka ʻōnaehana ma luna o nā manaʻo maikaʻi ʻole a i ʻole ka ʻikepili maikaʻi ʻole, akā paʻakikī ke ʻike, a e hana nā poʻe e hoʻohana ana i ka ʻōnaehana i kāna mea e haʻi ai iā lākou me ka nīnau ʻole. Nui nā ʻōlelo hoʻohenehene kahiko e pili ana i ka poʻe ʻauhau i kuhi hewa i kou inoa, a ʻoi aku ka paʻakikī o ka hoʻohuli ʻana iā lākou e hoʻoponopono i ka hewa ma mua o ka hoʻololi maoli ʻana i kou inoa. Nui nā ala e noʻonoʻo ai e pili ana i kēia, akā ʻaʻole maopopo i ka mea ʻoi aku ka maikaʻi: ma ke ʻano he pilikia ʻenehana ma SQL, a i ʻole he bug i kahi hoʻokuʻu Oracle, a i ʻole he hemahema o nā keʻena bureaucratic? Pehea ka paʻakikī o ka loaʻa ʻana o kahi bug i kahi kaʻina i alakaʻi i ka ʻōnaehana ʻaʻohe hiʻohiʻona hoʻoponopono typo? Ua ʻike ʻia paha kēia ma mua o ka hoʻomaka ʻana o ka poʻe e hoʻopiʻi?

Hōʻike ʻia kēia pilikia e nā moʻolelo i ka wā e holo ai nā mea hoʻokele i nā kahawai ma muli o ka ʻikepili kahiko o ka mea hoʻokele. ʻAe, pono e hoʻonui mau ʻia nā palapala ʻāina. Akā, pehea ka hewa o TomTom no ka puhi ʻia ʻana o kāu kaʻa i ke kai?

ʻO ke kumu aʻu i ʻōlelo ai ʻo ia, ʻo ia, e hoʻokumu ka bias aʻo mīkini i nā pilikia. Akā, e like kēia mau pilikia me nā pilikia a mākou i ʻike ai i ka wā ma mua, a hiki ke ʻike ʻia a hoʻoponopono ʻia (a ʻaʻole paha) e like me kā mākou i hiki ai i ka wā ma mua. No laila, ʻo kahi hiʻohiʻona kahi e hoʻopilikia ai ʻo AI i ka poʻe noiʻi nui e hana ana i kahi hui nui. ʻO ka mea maʻamau, e kākau kekahi mea ʻenehana ʻenehana a i ʻole mea kūʻai aku i kekahi mea ma ko lākou kuli, me ka hoʻohana ʻana i nā ʻāpana kumu wehe, nā hale waihona puke a me nā mea hana a lākou i hoʻomaopopo ʻole ai. A e kūʻai ka mea kūʻai maikaʻi ʻole i ka huaʻōlelo "artificial intelligence" i ka wehewehe huahana a, me ka ʻole o ka nīnau ʻana i nā nīnau, e hāʻawi iā ia i kāna mau limahana uku haʻahaʻa, e kauoha iā lākou e hana i ka mea a ka AI. ʻO kēia ka mea i hana ʻia me nā waihona. ʻAʻole kēia he pilikia naʻauao hana, a i ʻole pilikia polokalamu. ʻO kēia ke kumu kanaka.

hopena

Hiki iā ʻoe ke aʻo i ka mīkini i nā mea a pau āu e aʻo ai i ka ʻīlio - akā ʻaʻole hiki iā ʻoe ke maopopo i kāu mea i aʻo ai i ka ʻīlio.

Manaʻo pinepine wau e like me ka huaʻōlelo "artificial intelligence" wale nō i ke ala o nā kamaʻilio e like me kēia. Hāʻawi kēia huaʻōlelo i ka manaʻo hoʻopunipuni a mākou i hana maoli ai - ʻo kēia naʻauao. Ke hele nei mākou i HAL9000 a i ʻole Skynet - kahi mea maoli maopopo. Akā ʻaʻole. He mau mīkini wale nō kēia, a ʻoi aku ka pololei o ka hoʻohālikelike ʻana iā lākou me ka mīkini holoi holoi. ʻOi aku ka maikaʻi o ka holoi lole ma mua o ke kanaka, akā inā ʻoe e hoʻokomo i nā kīʻaha i loko ona ma kahi o ka holoi lole, e holoi ʻo ia. E maʻemaʻe nō nā kīʻaha. Akā ʻaʻole kēia ka mea āu i manaʻo ai, a ʻaʻole e hiki mai kēia no ka mea he manaʻo ʻino ka ʻōnaehana e pili ana i nā kīʻaha. ʻAʻole ʻike ka mīkini holoi i ke ʻano o nā kīʻaha a me nā lole - he laʻana wale nō ia o ka automation, ʻaʻohe manaʻo ʻokoʻa mai ke ʻano o nā kaʻina hana ma mua.

Inā mākou e kamaʻilio e pili ana i nā kaʻa, nā mokulele, a i ʻole nā ​​ʻikepili, e ikaika loa kēia mau ʻōnaehana a i kaupalena ʻia. E hilinaʻi lākou i ka hoʻohana ʻana o nā kānaka i kēia mau ʻōnaehana, inā he maikaʻi a maikaʻi ʻole paha ko lākou manaʻo, a pehea ka maopopo o kā lākou hana.

No laila, ʻo ka ʻōlelo ʻana "ʻo ka makemakika ka naʻauao, no laila ʻaʻole hiki ke loaʻa nā manaʻo" he wahaheʻe loa. Akā ʻaʻole like ka hewa i ka ʻōlelo ʻana he "subjective in nature." Loaʻa ka ʻike mīkini i nā mamana i loko o ka ʻikepili, a ʻo nā hiʻohiʻona i loaʻa iā ia ma muli o ka ʻikepili, a pili ka ʻikepili iā mākou. E like me kā mākou e hana ai me lākou. ʻOi aku ka maikaʻi o ke aʻo ʻana i ka mīkini ma mua o ka hiki iā mākou - akā ʻoi aku ka maikaʻi o nā ʻīlio ma mua o nā kānaka i ka ʻike ʻana i nā lāʻau lapaʻau, ʻaʻole ia he kumu e hoʻohana ai iā lākou i mea hōʻike a hoʻoholo i kā lākou hōʻike. A ʻo nā ʻīlio, ma ke ala, ʻoi aku ka naʻauao ma mua o nā ʻōnaehana aʻo mīkini.

Unuhi: Diana Letskaya.
Hoʻoponopono: Aleksey Ivanov.
Kaiāulu: @PonchikNews.

Source: www.habr.com