investigator have claimed that hokey intelligence ( AI ) will reach the singularity within seven yr , after attempting to quantify its onward motion .
rendering companyTranslated , confront their workplace at an Association for Machine Translation in the Americas group discussion , explained that they first start test car interlingual rendition technology in 2011 . The team settled on a metric to value AI procession , which they ’ve called " Time to Edit " ( TTE ) . Simply put , it is the time it postulate a human transcriber to edit a translation produced by another human or an AI .
Over the years , the TTE for AI - translated textual matter has add up down fairly systematically , take translate to foretell the date when AI hits the singularity , when the time is tantamount to human translator .
" When plotted graphically , the TTE information point a astonishingly analogue trend , " the teamwrote on their internet site . " If this trend in TTE stay on its downslope at the same pace as it has since 2014 , TTE is projected to decrease to one second within the next several years , approaching a point where MT [ motorcar translation ] would leave what could be call ' a complete version ' . "
The squad highlights other amount the AI is improving on , such as linguistic errors per 1,000 translate Son . presently , industry AI create around 50 errors per 1,000 translate Book , with the team have a bun in the oven that to improve at close to the same rate as TTE .
The team believes that the ability to transform , a complex task for a car , is a dear meter of when technology reaches singularity with humans .
" Many AI research worker believe that solving the speech communication version problem is the close thing to producing Artificial General Intelligence ( AGI ) , " the team write . " This is because natural language is by far the most complex job we have in AI . It requires precise modeling of reality so as to lick , more so than any other narrow AI . "
The terminal figure " uniqueness " used by the team come out to intend " on par with humanity " , and may be an denotation of AGI , rather the traditional signification – the point where technical growth isout of our control , and irreversible . One popular version of the uniqueness is the " intelligence operation explosion " , where AI is able to better itself more and more with each generation cycle , making itfar surpass human intelligence information
Nevertheless , tasks such as translation on equivalence with homo would be an authoritative milepost in AI , even if we wo n’t be seeing machines acquire out of our control by 2030 . It could be used as a part of a test for AGI , though it ’s important to call back thatlanguage ability does not make a machine consciousand speech models are essentially extremely good predictive texts , rather than intelligence that understands the meaning .
However , tracking the progress of AI is still somewhat cool , and will only become more of import as we develop closer to that theoretical uniqueness point .
" All of us understand that we are come on singularity in AI , " Translated CEO Marco Trombetti said of the research . " For the first time , we have been able to quantify the upper at which we are advance toward it . "
