Exploring the Frontiers of AI: Challenges and Limitations of ChatGPT

I guess it’s not necessary to introduce Chat GPT! We all have been tempted to use it and everyday we discover a new way on how this AI tecnology based on LLMs can make our life easier.

I though it might be interesting to present the other face of the coin here and dedicate a few minutes to think about the limitations and challenges of AI in general and Chat GPT in particular:

1. Ownership of responses can be unclear what can drive to copyright problems.
2. Bias on data is not new. Training data will potentially contain biases so this is something to keep on mind when using this technology. When thinking on the use you’ll do to your AI system, considering this limitation is a must if you want to make sure you are really evolving your decision making process.?Noways everyone knows this, but it’s never late to remember that AI learns from the past; so if you really want to boil the future of your company/project make sure your AI application takes this in consideration.
3. Losing track of the context in complex conversations
4. Hallucination. Yes, that’s correct! ChatGPT has a great imagination!

We can prevent this limitation by learning how to write better prompts or being realistic on the expectations we have on the system. This will help you avoid generating text that’s beyond your expectations or that contains harmful content.

5. Garbage in, garbage out: if the prompt (direction of the user) provided doesn't contain sufficient context or isn't written in an easily-interpretable way, chances are it will lower the quality of the response.

Talking about limitations is clear that leaving those aside, the AI technology is evolving at light speed and it’s shaping a new type of society.?The latest public request to Pause AI progress for 6 months that was punished recently and signed by people such as Elon Musk , Steve Wozniak, or Craig Peters is just a good example of the velocity on the improvement on AI.

Should we let machines flood our information channels with propaganda and untruth??Should we automate away all the jobs, including the fulfilling ones? …

This letter is echoing some of the limitations and challenges mentioned on this article, such as the biases on the AI systems or the hallucination syndrome of Artificial Intelligence.

I’m normally the first one to defend all type of evolution; however while some people still believing on the ¨good intentions of the human race¨, I believe that nobody here will deny the need for more robust governance systems or a more safe and robust public funding for technical AI safety research.

Bill Gates also have published his thoughts on his latest article The Age of AI has begun, where he focus on the different ways how AI can help to improve some of the most inequalities of the word.

It’s also clear for me the impact of AI in society and the potential to change us.. hopefully that change will be to better, but won't hurt to work to make sure of that! :)

And for who has read until here, in case you are asking yourself; no, I didn’t use Chat GPT to write this article so let me a comment here and it’d be great to read your thoughts!