Ever wondered if the latest NeuroCyber AI model named as C-3PO comes close to how people talk, think, and acquire knowledge. The model is no match for the original C-3PO’s fluency in 6 million forms of communication, but is replete with linguistic expertise and can learn from humans as humans do – from observation and imitation.
Unlike the earlier AI assistants such as Alexa or Siri that could recognize commands and act to perform the command, NeuroCybers can evolve into instinctive companions and assistants. One could be bought for Grandma for the holiday season.
NeuroCyber C-3PO is not the buy anytime soon, but construction of such intelligent helpers and pals may be possible if their learning abilities are engineered similar to those of humans or animals.
At present, some AI models are marching toward natural language processing using deep learning approach, which involves to mimic how humans acquire knowledge.
Meanwhile, GPT-3 – a language learning model from OpenAI is like a supercharged version of the autocomplete version on the phone.
From its definition, it can use a made-up word in a sentence, rewrite a paragraph in the way a famous author does. The assistant can write creative fiction, or write a code for a program based on details of its function. Furthermore, the assistant can answer queries about general knowledge.
Nonetheless, it is an overstatement to say GPT-3 can think like humans because it still makes silly errors that humans do not. In an article, GPT-3 penned that if one drinks a glass of cranberry juice with a spoon of grape juice will lead to death. It is impressive what GPT-3 can do, but it still needs to learn some basic facts of life. Thus, it is important to focus on what is missing.