The Language Revolution: Can AI Machine replace Human brain?

Dylan Yap Jin Quan
3 min readOct 22, 2023

--

AI image generated by Picsart

When we first come to this topic, the first impression is that “Are we going to be jobless or replaced by AI soon?” The answer is yes and no? Why?

Karim Lakhani, a professor from Harvard Business School express his impression “AI Won’t Replace Humans — But Humans With AI Will Replace Humans Without AI” (Harvard Business Review — https://hbr.org/). I’m totally agree with his points. AI itself has not enough ability to replace human force, as AI at least for now, should be under human supervision in the daily task. AI technology now is behaving like an assistant tool to most of the professions. For example, a graphic designer is now making use of the power of Generative AI (Mid journey AI) to minimise their time on creating art feature under a minute.

With the recent fast growing Large Language model after the release of ChatGPT, from LLama-2 by Meta AI being open source to the fine-tuned small language model by public contribution in Hugging-face, everyone now can have their own AI assistant being built in their own machine. And the introduction of Retrieval Augmented Generation (RAG) technologies, meaning that we can manage the AI assistant’s knowledge base. The research on chain of thoughts which can mimic the way of how complex problem being solved using human brain. All of the research directions are towards developing a better AI machine which are specifically trained for certain professions.

Moreover, Microsoft introduced a mind blowing state of art framework for AI agent : AutoGen. First of all, we have a brief overview on the evolution of Large Language model (LLM).
1. First level : LLM — A language model which can answer questions but have no memory.
2. Second level : AI agent — A language model which can answer questions with knowledge database and with memory.
3. Third level : Multi agent — A group of AI agents with their own professional knowledges to solve problem together.

Multi Agent > AI Agent > LLM.

Now, we can train or fine tune a single agent only focusing on certain tasks. Imagine we have a group of trained AI professions work together as a workforce. There are endless possibilities. The image below shows one github project named ChatDev as a virtual software company with different intelligent agents holding different roles. The url is available in the image caption.

Image from https://github.com/OpenBMB/ChatDev

Back to the question “Can AI machine replace human brain?”, I remain the question opened. Please feel free to discuss the question and leave your comments below.

--

--

Dylan Yap Jin Quan

AI Engineer | Data Scientist | LLM/RAG I will mostly focus on sharing LLM related topic which is an awesome game changer.