Masanin Shawara Wanda Ya Tattara AI: Yadda Injin Tunanin Kirsimeti na Herbert Simon Ya Canza Komai

Janairu 1956. Carnegie Institute of Technology. Wani malami mai shekaru 39 — wanda ya kware a fannin kimiyyar siyasa, wanda ya riga ya shahara a tsakanin masana tattalin arziki saboda ya yi ikirarin cewa babu wanda ke inganta komai — ya bude ajin sa na samfurin lissafi da wani jumla da ba ta dace ta zama gaskiya ba: A lokacin Kirsimeti, Al Newell da ni mun kirkiro wata na'ura mai tunani.

Injin ba ya wanzu sosai a matsayin injin. A lokacin hutu, Simon, Allen Newell da mai shirya shirin Cliff Shaw sun tsara Logic Theorist — kuma kafin ya taɓa gudana a kan kwamfuta, sun yi simintin hannu: an rubuta ƙananan shirye-shiryen shirin a kan katunan 3×5, an raba su ga iyalin Simon da ɗalibai, kowanne mutum yana aiwatar da dokokin wani ɓangare. Injin tunani, wanda aka fara gudanarwa a kan mutane. A lokacin bazarar, ya gudana a kan kayan aiki kuma an gabatar da shi a taron Dartmouth — taron da ya ba da sunan fasahar kere-kere.

Labari yana yawan ajiye shi a ƙarƙashin tarihin kwamfuta. Hakanan yana da alaƙa da kimiyyar yanke shawara — saboda Logic Theorist ba mai lissafi mai sauri bane. Wannan ra'ayin Simon na iyakar hankali ne da aka aiwatar a cikin lambar: idan hankalin ɗan adam yana aiki ta hanyar hanyoyi masu sauƙi da bincike na zaɓi maimakon lissafi mai zurfi, to, na'ura mai amfani da hanyoyi masu sauƙi da bincike na zaɓi na iya zama mai hankali ma. Duk abin da Simon ya gina a cikin AI yana bin wannan motsi guda.

CV da bai kamata ya wanzu ba

Herbert Alexander Simon (1916–2001) ya karɓi digiri na PhD a fannin kimiyyar siyasa (Chicago, 1943), ya rubuta Administrative Behavior (1947) akan yadda ƙungiyoyi ke yanke shawara a zahiri, kuma ya shafe shekaru guda hamsin a Carnegie Mellon yana aiki a fannonin tattalin arziki, ilimin halayyar dan adam, kimiyyar kwamfuta da falsafar kimiyya. Kyaututtukan suna nuna fadin: 1975 ACM Turing Award tare da Allen Newell, don gudummawar asali ga basirar wucin gadi da ilimin halayyar dan adam; 1978 Nobel Memorial Prize in Economics, don bincike na farko a cikin yanke shawara a cikin ƙungiyoyin tattalin arziki; 1986 U.S. National Medal of Science. Har yanzu yana ɗaya daga cikin mutane kaɗan da suka taɓa riƙe duka Turing da Nobel — don abin da ya dage cewa shiri ne guda na bincike.

Wannan shirin: fahimci basira kamar yadda take aiki a zahiri — a cikin mutane, a cikin kungiyoyi, a cikin na'urori — a ƙarƙashin iyakokin gaskiya na bayani, lissafi da lokaci. Rabon's ka'idar ta yi ritaya daga mutum mai hankali sosai na tattalin arziki. Rabon injiniya ya tambayi tambayar da ta dace: idan basira tana cikin bincike mai iyaka, shin za mu iya gina ta?

Mai Tunani na Lissafi: Shaida ta Hanyar Gajere

Mai nazarin lissafi ya kai hari kan ka'idojin Whitehead da Russell na Principia Mathematica — ginin lissafi na farkon karni — kuma ya tabbatar da 38 daga cikin 52 na farko a babi na biyu. Hanyar ita ce saƙon: maimakon yin aiki da kowanne fassara (karfi mai rauni wanda kayan aikin wannan zamani ba za su iya jurewa ba), ya yi amfani da heuristics don zaɓar hanyoyi masu yiwuwa, ya yi aiki daga baya daga manufofi, kuma ya tsaya lokacin da ya sami hujja mai inganci maimakon mafi kyau. Binciken heuristics da gamsuwa — kimiyyar yanke shawara ta Simon, tana gudana da saurin na'ura.

Don Theorem 2.85 an sami hujja mafi kyau fiye da asalin. Bertrand Russell, wanda a lokacin yana cikin shekaru tamanin, ya amsa da farin ciki lokacin da Simon ya rubuta masa game da shi. Journal of Symbolic Logic ya fi wahala a jawo hankalinsa: ya ki wallafa sakamakon, yana hukunta sabon hujja na wani ka'idar asali a matsayin wanda ba ya cancanci wallafa — a bayyane ba tare da lura da cikakken bayani cewa ɗaya daga cikin marubutan shine shirin kwamfuta ba.

Mai Magance Matsaloli na Gaba ɗaya: Tsari Ba Tare da Abun Ciki Ba

Shirin magaji ya kasance mai ƙwazo fiye da yadda za a yi tsammani daga marubucin Administrative Behavior. Mai Magance Matsaloli na Gaba ɗaya — sigar farko tana gudana a 1957, rahoton an buga shi a 1959 ta Newell, Shaw da Simon — ya gabatar da nazarin hanyoyi-da-makoma: auna bambanci tsakanin halin yanzu da halin burin, nemo mai aiki wanda ke rage wannan bambancin, yi amfani da shi, maimaita. Kewayawa ta hanyar rufe gibin.

Gaskiyar sabuwar fasahar GPS ita ce ta gine-gine: ta raba dabarun warware matsala daga abun cikin matsalar. Wannan injin guda zai iya kaiwa ga warwarewa, shaidar ko tsare-tsare, idan an ba da bayanin halaye da masu aiki. Wannan rabuwa — hanyar gaba ɗaya, yankin da za a haɗa — ta zama babban ka'idar zane na AI, kuma nazarin hanyoyi da manufofi har yanzu ana iya ganinsa a cikin tsarin shiryawa na zamani.

Babban Ikirari: Alamu da Bincike

A cikin jawabin su na kyautar Turing na 1976, Kimiyyar Kwamfuta a matsayin Bincike na Gaskiya, Newell da Simon sun taƙaita shekaru ashirin na aiki cikin shahararren zato na fannin: sistem alamu na jiki yana da hanyoyin da suka dace da isassu don aikatawa mai hankali gaba ɗaya. Hankali yana sarrafa alamu; kwamfutoci suna sarrafa alamu; don haka, idan an tsara su yadda ya kamata, kwamfutoci na iya aikatawa da hankali — kuma babu wani abu fiye da sarrafa alamu da ake buƙata. Wannan ya zama tsarin abin da yanzu ake kira AI na alamu, ko GOFAI.

Hakanan sun yi hasashe tare da kwanan wata a kansu, kuma kwanakin sun kasance ba daidai ba. A cikin takardarsu ta 1958 Operations Research, Simon da Newell sun yi hasashen cewa cikin shekaru goma kwamfutar dijital za ta zama zakaran chess na duniya, sai dai idan dokokin sun hana ta shiga gasar. Zakaran duniya ya faɗi ga kwamfuta a 1997 — shekaru 39 bayan hasashen, ba shekaru goma ba. Abu ne mai kyau a lura da irin kuskuren da wannan yake: tsarin (binciken heuristics a cikin sarari wanda kowace na'ura ba za ta iya cika ba) ya kasance daidai; lokacin ya yi kuskure da wani zamani. Dukkanin abubuwan suna cikin tarihin.

Shin Deep Learning ba ta tabbatar da cewa Simon ba daidai bane?

Ga korafin da mai karatu na 2026 ya kawo: AI na zamani ba alama bane. Hanyoyin sadarwa na jijiya suna koyon tsarin daga bayanai; babu wanda ke rubuta ka'idojin hannu; kuma rubutun Richard Sutton na 2019 mai tasiri The Bitter Lesson yana karantawa kamar hukunci kan dukkan hanyar Simon — shekaru saba'in na tarihin AI suna nuna cewa hanyoyin gaba ɗaya da suka dace da lissafi sun fi tsarin da aka gina akan ilimin fannin da mutum ya ƙirƙira.

Hujjar ta yi nasara — a wani ɓangare. Babban ikirarin alama ba ya tsira: ya bayyana, alamu ba su isa ba, kuma shirye-shiryen ilimi da aka gina da hannu na shekarun 1970 da 80 sun tsaya a inda Sutton ya bayyana. Bayanin gaskiya na Simon ya bayyana hakan a fili. Amma karanta kammalawar The Bitter Lesson: iyalan hanyoyin guda biyu da suka yi fice sune bincike da koyo. Binciken heuristik shine gudummawar Simon da Newell ga fannin, yana nan a yau daga binciken itacen wasa zuwa juyin tunani na gangan na tsarin AI na yanzu. Kuma abin da manyan hanyoyi ke yi — gane tsarin da aka tara daga gagarumin kwarewa — yana da kusan kusanci da ma'anar kwarewa ta Simon. Koyon karfafa, wanda masu kafa Andrew Barto da Richard Sutton suka karɓi kyautar Turing ta 2024 (da aka sanar a Maris 2025), bincike ne na zaɓi na gwaji da kuskure a ƙarƙashin iyakokin lissafi — shirin bincike da Simon zai gane a matsayin dangin.

Don haka, an raba sakamakon, kuma rabon yana da ma'ana: ginshiƙi da ya dogara da shi ya sha kashi; ka'idar hankali da ke ƙarƙashinsa ta yi nasara. Na'urori ba su zama masu hankali ta hanyar lissafi mai zurfi ba. Sun zama masu hankali ta hanyar bincike mai iyaka, mai amfani da dabaru, wanda ke motsawa ta hanyar ganowa — wanda shine ikirarin da Simon ya dogara da shi a cikin aikinsa. Ayyukan da suka gabata suna rufe zagaye daga wani ɓangare: binciken 2025 ya gano manyan samfuran harshe suna nuna iyakacin hankali da kansu, suna kaucewa daga ingancin wasan kwaikwayo a hanyoyi masu kama da na mutum, yayin da aka tsara gamsuwa a cikin daidaiton samfurin a matsayin ƙuntatawa na matakin burin.

Fingernan Simon akan AI na Zamani

Lokaci: Rayuwar Wani Masani Mai Ilimi da Dama

Abin da za a ɗauka daga Simon cikin zamanin AI

Hanyoyi hudu na aiki ga duk wanda ke gina tare da, ko yanke shawara tare da, na'urorin da aka iyakance:

Inda Argumentree Ya Dace

Simon ya dauki hankali na mutum da na na'ura a matsayin batun guda: wakilan da aka iyakance suna bincike a ƙarƙashin ƙuntatawa. Argumentree an gina shi bisa wannan tunani. Huɗa na mutum da huɗa da AI ta fitar suna sauka a cikin tsari guda mai duba — itacen — inda tunani ke bayyana, kimantawa suna jawo hankali mai ƙanƙanta ga mafi ƙarfi, kuma rikodin yana ci gaba a matsayin ƙwaƙwalwar ƙungiya.

Don ƙungiyoyin da ke buƙatar a duba amfani da AI, samfurin 'yar uwa AIAgentree yana faɗaɗa wannan ra'ayi zuwa ga tunanin inji da kansa — bin diddigin yadda tsarin AI ya kai ga ƙarshe, a cikin tsarin da Simon zai tambayi a gani.

Gagarumin Kwatancen

Agogon catur Simon ya yi hanzari sau hudu, tsarin alama ya rushe, kuma filin da ya kafa tare da wasu yanzu yana gudana bisa hanyoyin da ba ya gina. An yi masa hukunci a matsayin annabi, tarihin hadaddiyar. An yi masa hukunci a matsayin kimiyya, wani abu mai wuya: babban ikirari — cewa basira, ko ina take faruwa, tana cikin bincike mai iyaka wanda ke jagorantar ta da ganewa — ya wuce kowanne tsarin da aka yi amfani da shi wajen gwadawa, ciki har da nasa.

Wannan shine dalilin da ya sa jimlar aji har yanzu tana da tasiri bayan shekaru saba'in. Ba ya sanar da wani na'ura. Ya sanar da cewa tunani ya zama wani fannin injiniya — ga kwakwalwa, ga kungiyoyi, kuma yanzu ga na'urori.