As you read this story, you will learn the following:
-
By a unique metric, we could reach technological singularity by the end of this decade (or even earlier).
-
One translation company has developed a metric, Time to Edit (TTE), that calculates how long it takes a professional human editor to fix an AI-generated translation compared to a human translator. This might help quantify the speed toward the singularity.
-
Artificial intelligence that can translate speech as well as humans can transform society.
In the world of artificial intelligence, the concept of “singularity” is particularly important. This elusive concept describes the moment when artificial intelligence transcends human control and rapidly transforms society. The tricky thing about the AI singularity (and why it borrows terminology from black hole physics) is that it’s hard to predict where it will start, and almost impossible to know what lies beyond this technological “event horizon.”
However, some AI researchers are looking for signs of reaching the singularity, measured by advances in AI approaching human skills and capabilities.
One such metric defined by Rome-based translation company Translated is the ability of AI to translate speech with human accuracy. Language is one of the most difficult artificial intelligence challenges, but computers that can close this gap could theoretically show signs of artificial general intelligence (AGI).
“That’s because language is the most natural thing for humans,” said Translated CEO Marco Trombetti. Said at a conference in 2022 In Orlando, Florida. “Nonetheless, the data collected by Translated clearly shows that machines are not far away from closing the gap.”
The company tracked AI performance from 2014 to 2022 using a metric called Time to Edit (TTE), which calculates how long it takes a professional human editor to fix an AI-generated translation compared to a human translator. During this 8-year period, more than 2 billion After post-editing, Translated’s AI showed slow but undeniable progress, slowly closing the gap with human translation quality.
Translated
According to Translated, on average, it takes a human translator about a second to edit each word of another human translator. In 2015, professional editors checking machine translation (MT) suggested that it took about 3.5 seconds per word, whereas today, that number is just 2 seconds. If this trend continues, Translated’s AI translations will be as good as human translators by the end of the decade (or even earlier).
“The change is so small that you don’t feel it every day, but when you see the progress … across 10 years, it’s impressive,” Trombetti said on podcast. “This is the first time anyone in the field of artificial intelligence has predicted the speed of the singularity.”
While this is a novel way to quantify humanity’s proximity to the singularity, the definition of such a singularity suffers from similar problems Identify AGI more broadly. While perfecting human speech is undoubtedly at the forefront of AI research, impressive skills don’t necessarily make a machine intelligent (not to mention how many researchers don’t even agree What is “intelligence”).
Whether or not these ultra-accurate translators herald our technological doom, that doesn’t diminish Translated’s AI achievements. Even if a true “technological singularity” remains elusive, artificial intelligence capable of translating speech as well as humans could very well transform society.
You may also like
