This AI Translation Firm Has a Timeline to the Singularity You Won't Believe

 

AI translation firm unveils a new method to measure progress towards singularity: Translated, an Italian company that specializes in translation services, has revealed a unique way to track advancements in AI by analyzing advancements in machine translation. Using this method, the company aims to predict when the technological singularity, a theoretical point at which machines surpass human intelligence, will be achieved.

AI translation firm sets milestones for singularity: Translated, a Rome-based company, has established a benchmark for measuring progress towards the singularity. According to the company's new research, the singularity will be achieved when AI provides a "perfect translation" - when machine translation (MT) surpasses the quality of the best human translations.

Translated, a Rome-based company has used its analysis of advancements in machine translation to predict when the technological singularity will be achieved. According to the company's CEO, Marco Trombetti, the singularity will occur within the 2020s, at least for the top 10 languages in average complexity. Trombetti added that in some specific domains and languages, this point has already been reached, while in others it may never be achieved.

AI translation firm's predictions based on data from Matecat: Translated, a Rome-based company, has used data from Matecat, a computer-assisted translation (CAT) tool, to estimate when the technological singularity will be achieved. Mate, which began as an EU-funded research project in 2011, was later released as open-source software that professionals use to improve their translations. Translated predictions are based on data taken from the platform.

Translated uses Matecat to track progress towards singularity: Translated, an Italian company, uses data from its freemium product, Matecat, a computer-assisted translation tool, to estimate when the technological singularity will be achieved. Users provide data to improve the company's models in return for using the tool. To chart, the progress towards singularity, Translated tracked the time spent by users checking and correcting 2 billion machine translation suggestions, made by around 136,000 professionals worldwide, across Matecat's 12 years of operation. 

The translations spanned diverse domains, from literature to technical subjects, including fields in which machine translation is still facing challenges, such as speech transcription.

Data suggests rapid improvement in AI: Translated data from Matecat, its computer-assisted translation tool, suggests that AI is making rapid progress towards singularity. The analysis of time spent by professional translators checking and correcting machine translation suggestions over 12 years of operation shows a decrease in the average time per word. In 2015, it took around 3.5 seconds for world-leading translators to check and correct MT suggestions, while today it takes around 2 seconds per word.

Rapid improvement in AI suggests "perfect translation" will be achieved soon: Translated data from Matecat, the computer-assisted translation tool, shows that AI is making rapid progress towards singularity. At the current rate, the time spent by professional translators checking and correcting machine translation suggestions will hit 1 second in around five years, at which point machine translation (MT) will provide the epochal "perfect translation." This means that in practical terms, it will be more convenient to edit a machine's translations than a top professional's. 

According to Marco Trombetti, CEO of Translated, any task involving communication, understanding, listening, and sharing knowledge will become multilingual with minimal investment. He states that the exact date of when we will reach the singularity point may vary, but the trend is clear: it is really close.

Linear progress in MT leads to increased demand: 

Translated, a Rome-based company has used data from its computer-assisted translation tool, Matecat, to predict when the technological singularity will be achieved. The research shows a highly linear rate of development in machine translation (MT) which is surprising as it was believed that progress would slow as singularity approached. 

Advances in MT require increasing computing power, linguistic data, and algorithmic efficiency. If this momentum continues as predicted, Translated anticipates a significant increase in demand for MT, at least 100 times higher. While some workers may worry that their jobs will be automated, the company also forecasts at least a tenfold increase in requests for professional translations.

Advances in MT will lead to increased demand for human translation:

 Translated, an Italian company, predicts that even as machine translation (MT) becomes more advanced and efficient, there will still be a need for human translation. According to the company's CEO, Marco Trombetti, all customers who are using MT on a large scale are also spending more on human translation. He believes that MT is an enabler that creates more interactions between markets and users that were not in contact before. This generates business, and the business generates higher-quality content that requires professionals. Trombetti also expects new roles to emerge for elite translators as a result of the increased demand for human translation.

The CEO of the company, Marco Trombetti, states that to achieve the best quality from MT, it needs to be trained by the best linguists. As a significant volume of translations is required to train language models and fix errors, he believes that there will be huge competition for the best translators in the upcoming years.

The first to ever quantify the speed at which we're approaching singularity by using the improvements in machine translation (MT) as a barometer. The company claims that the data they collected from their translation tool Matecat, suggest that the singularity point will be achieved in the 2020s. The use of MT as a measure of AI progress is considered compelling as human languages are notoriously tricky for machines to master. The subjectivity of linguistic meaning, the constantly evolving conventions, and the nuances of cultural references, wordplay, and tone can be elusive for computers.

According to the company's CEO, Marco Trombetti, MT is a good predictor of what is coming next in AI. This is because, in translation, the complexities of human languages must be modeled and linked in two languages. As a result, the field of machine translation is often pioneers in algorithmic research, data collection, and model sizes. For example, the Transformer model was applied to MT many years before being used in OpenAI's GPT systems.

The Italian entrepreneur is anticipating a new era for global communication with the advent of singularity. He envisions universal translators and all content becoming globally available, which would allow everyone to speak their native language.