Artificial Intelligence is this year’s hot topic (the way Crypto was a couple of years ago, and we all saw how that ended). I know it would be wrong to compare AI with Crypto, as the former has some real-world utility. I am just referring to the hype similarity between AI and crypto.
AI, or Generative AI to be specific, become the topic of town since late January 2023, when Microsoft announced its $10bn investment in ChatGPT.
I was recently speaking with a couple of senior IT professionals, and they strongly believed that AI would improve over the years and could train itself and, over time, replace the human brain. One even argued that AI would be better than the human brain. This belief has a major flaw; let me explain this with an example. AI can very well replace some repetitive human tasks, similar to how automation replaced factory workers. AI can also help humans increase speed and efficiency in completing creative tasks; however, there is no way it can replace the human brain or touch everything we do.
The main issue with Gen AI is that it cannot interpret things; rather, it will never have that ability. AI depends on the data at its disposal to provide the output. The AI output is as good as the input it receives. Someone means humans must create this data or information for AI to process. AI can not generate information on its own or has no imaginative abilities. It may be able to generate some basic information. However, there is no way AI can have a deeper understanding of a subject the way the human brain can. This is common sense if you know how these Gen AI tools work.
I recently tested paid versions of three leading-gen AI tools for a month. When writing my blog about the collapse of the Silicon Valley Bank, I tried to generate output from these AI tools to check what answer I got. All these tools gave the typical high-level answers that were widely discussed in the media. However, it was not even 10% close to the detailed analysis I have in my blog or the analysis conducted by some leading industry practitioners or academic scholars on this topic. For example, in my blog, I outlined in detail how a specific US Liquidity Coverage Ratio regulation called LRMS (Liquid and Readily Marketable Standards) was important, or Yale University or Bank Policy Institute publishes a detailed analysis with their view on the SVBs LCR ratio. There is no way a Gen AI tool can perform such analysis independently. The human brain is unique, perfectly explained in this Harvard Business Review article, a specific snippet below for your quick reference.

I was disappointed with my experience with these Gen AI tools to get any good output for the issues that involve SME knowledge. Where I found these Gen AI tools, some decent output was for the informative posts. For example, you will get a decent response if you want detailed information about Acadia National Park and ask that query in ChatGPT. This type of query does not require a specific skill set to resolve or require no interpretation.
To conclude, I believe Gen AI will be useful and improve a lot in the future. It’s definitely going to change the way humans work. It may prove superior to humans in doing all the repetitive tasks. However, the current hype over it is not justifiable. AI can’t substitute the human brain and its ability to imagine without data to interpret complicated information where expert judgment is involved.
Update: I wrote this blog in August 2023, and a few months later, there was the whole Sam Altman firing saga at OpenAI. The simple point is, why was it so crucial for Sam Altman to continue as CEO at OpenAI? Why can’t OpenAI was not able to simply feed all his emails to ChatGPT and let the AI model make all the future decisions of the company? The fact is, it’s not that easy. Even the company that came up with ChatGPT can’t let go of its CEO, and all its employees decided to quit to get him back. It just highlights that even the makers of CHatGPT don’t believe in this over-the-top narrative that AI can replace humans. Yes, it can make humans more efficient, but that’s it.