My Corner of the Web
I just published a new article over at LinkedIn (more on that in a moment) entitled:
The subtitle is: Lessons Learned and 3 Laws for Generative AI!
This image is generated by Dall-E so there is no copyright. AIs cannot get a copyright (though they are being sued by Getty Images so things could change).
Why did I post that article to LinkedIn? Besides plumbing the depths of ChatGPT hallucinations it is basically a policy article and many of my Government colleagues are over on LinkedIn so it was a natural choice for that.
This article will be one of a series of articles. Here is how how that article is split into 3 parts:
This article is broken up into three parts:
1. The Problem – After an introduction to large language models (LLMs), I propose a use case to examine its hallucination problem by asking it about myself. I encourage you to do the same. If you are not recognized by ChatGPT, I recommend you try asking about someone you know well who may have more publicity on the internet for which the model may have been trained on.
2. Exploring Alternatives – Before suggesting a solution it is important to do some due diligence and explore what alternatives are being studied and pursued in the field. The 2nd part of this article will do that.
3. A proposed solution or the “Three Laws of Generative AI” – The final part of this series will propose a solution to the problem in the form of “Three Laws of Generative AI” modeled after Isaac Asimov’s “Three Laws of Robotics”.
Additionally, I have begun delving into writing code for simple neural networks to make sure I understand the underlying technical details. Maybe I will have ChatGPT teach me about neural networks ... I like the idea of a Neural Network teaching me about Neural Networks!
More on that coming soon...