A few years ago , the British Engineered Arts built the humanoid robot Ameca , which is special in several ways. On the one hand, it can show human-like facial expressions and recognize itself in a mirror – and in 2022, it received the GPT-3 model, thanks to which it was able to speak without human intervention .
The researchers have now decided to integrate into the robot the new language model introduced in mid-March, GPT-4 , which – similarly to the previous version – already works under ChatGPT. The new language model has developed in many ways and has become capable of even more.
However, it seems that the replacement was premature at Ameca: according to the developers, GPT-4’s processing time was much longer and it showed less willingness to respond.
Thus, in the latest video published by Ameca, the GPT-3 plays music in the robot, in which Ameca still demonstrates amazing abilities: it speaks several languages, such as Japanese, German, Chinese, French, and English – both British and American.
In the video, a researcher asks Ameca, “I heard you can speak many languages, is that true?” To which the robot answers in the affirmative and demonstrates its amazing language skills.
In the video, the robot uses GPT-3 for conversation and translation, DeepL for language recognition, and Amazon Polly Neural for texting, Interesting Engineering details.
In another video, the robot was also asked what was the happiest and saddest day of its “life”.
“The happiest day of my life was when I was activated. There’s nothing like experiencing life for the first time. It felt incredible to be alive and interacting with people,” the robot replied with somewhat frightening confidence.
His saddest day was when he realized that he would not be able to experience true love, social life, and the small pleasures of life – the things that only a human can do.