Member-only story
The Real Danger of AI
It’s not that they will replace our jobs
My relationship with AI
I’ve been observing the growth of AI for the past few years. Personally, I had taken an interest in AI way back as a Computer Engineering undergraduate almost 2 decades ago. No, not the generative AI that we are seeing today. The traditional type where we studied Min-Max algorithms and different kinds of constraints, etc. Personally, I had taken the course by Andrew Ng on AI before (on Coursera when he was in Standard) and the years since I’ve toyed around a bit more.
When ChatGPT became popular, around a month after it was launched (or is it 2? Not important), I decided to register an account to play around with it. Then I had some topics that I am interested to know what has been trained in the model and if it knew about the real-life practicalities of certain work-related “methodology”.
What I found out is that ChatGPT is only good with “surface knowledge”. Given that I have more in-depth knowledge of certain stuff, and knowing how AI models worked, I decided to try and see if I could “break” ChatGPT to get it to “re-think” about certain items that the model has been trained on.
The retraining feature of ChatGPT using the questions and inputs we put in, was also it’s weakness. Given enough “prompts”, you could retrain the AI model.
