Understand limitations (and consequences).
First, it’s important to understand how technology works to understand what exactly you’re doing with it.
ChatGPT is essentially a more powerful and sophisticated version of your phone’s predictive text system, using words learned from vast amounts of data collected from the web to complete sentences as you type. I suggest
Nor can I be sure that what I am saying is true.
When you code a program with a chatbot, you can find out how your code was compiled in the past. Code written in chatbots can be buggy or insecure because the code is constantly being updated to address security vulnerabilities, Christian said.
Similarly, if you’re using ChatGPT to write an essay on a classic book, chances are the bot will construct a seemingly plausible argument. But if other people publish an erroneous analysis of this book on the web, it may show up in your essay as well. If the essay is then posted online, it will contribute to the spread of misinformation.
Melanie Mitchell, an AI researcher at the Santa Fe Institute, said:
In other words, bots do not think independently. I can’t even count.
Case in point: When I asked ChatGPT to compose a haiku about the cold in San Francisco, I was dumbfounded. It spat out a line with an incorrect number of syllables.
Fog covers the city,
The cool breeze chills me to the bone,
San Francisco winter.
OpenAI, the company behind ChatGPT, declined to comment for this column.
Similarly, AI-powered image editing tools like Lensa use existing images on the web to train their algorithms. So if women were presented in a more sexual context, the machine would reproduce that bias, Mitchell said.