The concept of technological singularity, where AI surpasses human control and rapidly transforms society, is a topic of great interest in the field of artificial intelligence. While it can be challenging to predict when and how singularity will occur, some AI researchers are searching for signs of its approach by measuring AI progress in relation to human skills.
One metric that has been identified is an AI’s ability to translate speech with human-level accuracy. Language translation is considered one of the most difficult challenges for AI, and if an AI system can match human-level accuracy in translation, it could be a sign of Artificial General Intelligence (AGI) and an indication that technological singularity is on the horizon.
Translated, a Rome-based translation company, has been tracking its AI’s performance since 2014 using a metric called “Time to Edit” (TTE), which measures the time it takes for human editors to correct AI-generated translations compared to human translations. Over the eight-year period and analyzing over 2 billion post-edits, Translated’s AI has shown a steady improvement, slowly closing the gap towards human-level translation quality. This suggests that the possibility of technological singularity may not be far off.