AI and its Hype Men.

Large Language Models have come a long way. Do they presage the imminence of AGI?

Artificial Intelligence is being talked about so much that I am swamped with this information on Twitter. Whenever I check on my timeline I encounter at least one thread based on it. Perhaps I follow such interesting people who are bullish about it. Some even predict that we are just few years away from the oncoming of Artificial General Intelligence.

I read this interview published in IEEE Spectrum of on Rodney Brooks which gave me another, a more skeptical perspective about the imminence of AGI. It is a fun conversation to read. Brooks does not discount the benefits of LLMs but wants everyone to calm down as far as the hopes or fear of inevitable AGI is concerned. Rodney Brooks is an AI expert and researcher. He feels that the world is wildly overestimating Open AI’s Large Language Models.

Mr. Brooks argues that LLMs are quite “stupid”, for lack of a better word. He says that they are still far away from competing with humans at an intellectual level.

He went on to say that the likes of ChatGPT4 and Bard don’t have any underlying model of the world. It is just correlation with language. I may be guilty of paraphrasing him a bit. Read the interview. You will not regret it.

One interesting part that came up in the conversation was when Rodney Brooks tells the interview that when he asked ChatGPT as to who is Rodney Brooks. It returned the answer that he is an AI skeptic, to which Rodney agreed.

My perspective as a tech enthusiast is that many people have different understanding of Artificial General Intelligence. Reading the interview cleared many of my niggling doubts about the subject. It has certainly expanded the scope of my understanding about AI. All the hoopla surrounding AI, LLMs may induce us into thinking that these systems would become self-aware any moment now. We may be overly optimistic about the capabilities of these Large Language Models and overlook many of their limitations. When they churn out their output, it is us who ascribe meaning and significance to that output.

Source : https://spectrum.ieee.org/gpt-4-calm-down

Additional Read: https://futurism.com/the-byte/ai-expert-chatgpt-way-stupider