Artificial intelligence in translation within seven years balances people, research from Translated found. From there, it’s just a few steps to the same success in other fields. Scientists warn that AI can then get out of control. What will happen next and what the world will look like, no one knows for sure. The IFLScience server informed about it.
Are we one step away from artificial intelligence replacing human beings in society? Experts from the company Translated came up with a disturbing study. Using the Time to Edit (TTE) model, researchers tracked AI performance over time.
The model determines the time it takes for a human to edit a translation created by another translator or AI. Using TTE, experts tracked the AI translation trend from 2014 to 2022.
Artificial Intelligence May Hit The Singularity Within 7 Years, Researchers Claim https://t.co/I36Sq2SMBV #AGI #ArtificialIntelligence #singularity
— Future AI (@FutureAI_inc) January 26, 2023
Translated published the results of the study on your website together with a graph showing that AI will be equivalent to a human in translation within seven years. While human translators kept the same quality, TTE decreased linearly with AI.
“If this trend in TTE decline continues at the same rate since 2014, we predict that TTE will decrease to one second within a few years,” wrote a team of experts on their website.
Machine translation then achieves something that can be called a “perfect translation.” But there are other parameters that track how the AI learns, such as the number of errors per thousand words. Currently, machine translation makes about 50 errors per thousand words.
The point of no return
However, scientists expect progress even in error rates comparable to TTE. Artificial intelligence will subsequently become artificial general intelligence (AGI), that is, artificial intelligence that has surpassed the goals for which it was created.
In that case, there are only a few steps away from achieving the same successes in other fields. What happens next? No one knows, so scientists refer to this point as reaching the singularity, the website writes IFLScience.
We know the term singularity from the study of black holes. It describes a state where we reach a certain point of no return. If AI reaches that level, we can only imagine what will come next.
Scientists explain this concept in the field of AI with several scenarios. According to some opinions, machines will reach the level of humans, according to others, they will eclipse human intelligence and get out of control. They can then change the world with new research that we don’t understand. Or destroy the world.
Even without apocalyptic scenarios, however, increasingly advanced AI presents a challenge. For example, a highly advanced program ChatGPT by OpenAI led to US schools banning it due to concerns about student cheating. The chatbot will answer every question with surprising efficiency.
Apple left AI again narrate audiobooks. This is because it is their more practical and less expensive creation. However, writers and publishers criticize this because one gives added value to the work, which makes art art.
