Kimiyyar kwamfuta ba sau da yawa ba take samun irin wannan watan. An rushe daya daga cikin imaninta na asali, kuma wadanda ke duba lalacewar sun hada da masu binciken da suka gina ta.
A ranar 16 ga Oktoba, wani research model a OpenAI ya samar da hujja ta mahmada Exponential Time Hypothesis a cikin kalmomi na yau da kullum. A cikin kwana daya, an kwafe hujjar zuwa Lean, wani proof checker wanda ke tabbatar da kowane mataki na dabara ta hanyar inji. Takardar manuscript mai shafi 67 tana ambaton OpenAI da kanta a matsayin marubuci.
Anthropic, wadda ke aiki ba tare da wata hadin gwiwa ba, ta riga ta isa wannan matsayin awanni kadan da da farko. Takardarta mai shafi 49, "A 2^o(n) Time Algorithm for 3-SAT", an danganta ta ga Claude, kuma wani Claude research model ne ya gano algorithm ta asali a wannan ranar.
Wannan hypothesis ta samo asali ne daga kusan shekarar 1999, lokacin da Russell Impagliazzo da Ramamohan Paturi suka kawo hujjar cewa Boolean satisfiability ba ta da wata algorithm mafi sauri fiye da exponential time. Wannan ra'ayi guda daya yana goyon bayan daruruwan sakamakon conditional hardness a graph theory, databases, computational biology da geometry.
Kungiyoyin biyu yanzu suna bayanin hanya ta deterministic wadda ke kirga satisfying assignments a cikin subexponential time, kuma dukansu biyu sun isa kan wannan babban kashi a exponent. Idan masu bita sun tabbatar da aikin, daya daga cikin katangun da aka fi amincewa da su a fannin zai nuna cewa yana da kofa.
Abin da ke ba wannan ikira nauyi ba asalin model din ba ne, sai tsarin dubawa da ke kewaye da su. Kowace hujja tana wucewa ta Lean. Kuma kwamitin bita da OpenAI ta tara ya hada da Impagliazzo da Paturi da kansu, wadanda yanzu suka sami kansu suna duba rushewar conjecture tarsu. Stefan Szeider na TU Wien ya duba kayayyakin jama'a kuma yana sa ran za su tsaya kansu karatu mai zurfi.
Taka tsantsan har yanzu yana da muhimmanci. Sakamakon ba ya faɗin komai game da P versus NP, kuma sabon algorithm har yanzu yana aiki a cikin superpolynomial time, don haka babu wani practical solver da zai yi sauri gobe safe. Abin da ya sauya shi ne taswirar theoretical, da ko wanene - mutane ko systems - ke sake zana ta.
Ga manyan kamfanoni, darasin yana nesa da complexity classes. Software wanda kwanan nan ya ke taƙaita takardun yanzu yana samar da lissafi na ainihi wanda inji ya tabbatar. Iyakar abin da AI na iya bayar da gudummawa a bincike na gaske yana tafiya da sauri fiye da yadda mafi yawan zagayowar shirin suke karkatawa.
Kamfanonin da za su koyi yin aiki tare da irin wadannan systems, maimakon kawai su ba su aikin bugawa, su ne za su kafa saurin shekaru goma masu zuwa.
Yi tunani a lissafi. Yi aiki da wayo. Jagoranta da dabara. Gina da alhakin amfani. Fadada a duniya. Kirkira ainihin darajar kasuwanci ta hanyar basira wucin gadi.
