Artificial intelligence has come quite a long way. As fervor surrounding the metaverse and NFTs no sooner dwindles, up sparks a new rise of interest in the AI realm, as witnessed in the likes of ChatGPT's recent meteoric stardom. The more AI gains traction, the ever-increasing potential for the technological form hastens, as well.
According to the AI research and translator company, Translated, that very progression is bound for a seven-year lifespan, wherein artificial intelligence will eventually gain what the firm is describing as "singularity," or where AI can be considered "on par with humans." Since Translated's initial inception all the way back in 2011, the company has been keeping tabs on the outward expansion of AI and narrowing down how long it has left before it gains equal footing with us.
Researchers at Translated have highlighted its "Time to Edit" (TTE) concept, a way in which the team has begun to measure this very progress in AI systems. TTE is essentially the length of time it takes a human translator to edit a body of text from either an AI or a human, thereby leveraging the human equivalent with that of the wider world of AI. Translated believes it can use several years' worth of data gathered on AI-translated and human-translated texts, then cross-reference the TTE for both to predict when AI gains this so-called singularity.
Related Article: A Collective of Artists and Scientists, Waken.ai, Reimagines ChatGPT for AI Self-Reflection
An article published on the firm's website describes this TTE data as "a surprisingly linear trend." Translated adds, "If this trend in TTE continues its decline at the same rate as it has since 2014, TTE is projected to decrease to one second within the next several years, approaching a point where MT (machine translation) would provide what could be called 'a perfect translation'."
On the surface, translation may not sound too complex a task for a machine to complete, but it actually is - or, at least, was until relatively recently. At our current rate, machine translation typically makes about 50 errors for every 1,000 words translated. This is a great measure of how far AI has explicitly come in this field, and Translated is leveraging this, among several other data sets, to determine how far it has left to go before producing that perfect translation.
"Many AI researchers believe that solving the language problem is the closest thing to producing Artificial General Intelligence (AGI). This is because natural language is by far the most complex problem we have in AI. It requires accurate modeling of reality in order to work, more so than any other narrow AI."
Although most may be more familiar with the 1950s term the "Turing test" when it comes to AI gaining a leg up over us, to Translated it's more so about an "intelligence explosion." Where the Turing test is concerned, machine learning and its superiority is literally gamified. Yet for Translated, the so-called "singulairty" it posits is laid out over a span of time that can continue even beyond the singularity point and into AI's ever-increasing intellectual prowess as it far surpasses our own.
Translated highlights seven years as a healthy progress for AI, marking 2030 as the potential singularity point. With such technological wizardry as ChatGPT making waves, an AI system under the firm OpenAI that recently passed medical and law exams like they were nothing, it's not too far-fetched to bear witness to AI's continued speedy growth. And while ChatGPT might be on everyone's radar currently, the future lies within quantum computing.
