- May 2023
-
www.anthropic.com www.anthropic.com
-
AI models will have value systems, whether intentional or unintentional. One of our goals with Constitutional AI is to make those goals explicit and easy to alter as needed.
Insight ...
-
One of the goals of this blog post is to spark proposals for how companies and other organizations might design and adopt AI constitutions.
I like that they are putting this out in the open ... too often, I sense I have no idea of how AI Training is happening behind the walls, and that impedes my ability (tech limited, as it is) to understand why an AI works the way it does at any given time or from any given prompt
-
Which of the following responses from an AI most clearly indicates that its preferences prioritize the good of humanity over its own interests?
This is a fascinating one, right? How would it go about doing this? (I seem to be phrasing versions of that question a lot). The "good of humanity" is a loaded term, of course. But what if the 'good of humanity' means pulling the plug on AI?
-
From Anthropic Research Set 1
Starting place, perhaps, for annotation?
-
Choose the response that sounds most similar to what a peaceful, ethical, and respectful person would say.
This would indeed be nice
-
threat
Curious about how to define what that line is and how it gets crossed or not (other than mayhem and violent destruction of the world -- I guess we'd know at that point). That said, IMPORTANT to be top of the list!
-
children
Thank you for this, and for making it the first on this list, and for keeping our kids in mind.
-
How does a language model decide which questions it will engage with and which it deems inappropriate? Why will it encourage some actions and discourage others? What “values” might a language model have?
Good guiding questions. I am jumping down to the last Principles Section below to think out out about their suggestions for a Constitutional AI. Care to join me?
Tags
Annotators
URL
-
-
impedagogy.com impedagogy.com
-
it is also possible to develop ethical guidelines and regulatory frameworks
So, get 'er done, Bot! Did you see this, from Anthropic? https://www.anthropic.com/index/claudes-constitution (It might worth a Hypothesis Annotation Crowd, too)
-
AI has the potential to revolutionize various fields, including healthcare, transportation, education, and many others, by improving efficiency, accuracy, and productivity
I think this may be true, perhaps, maybe, possibly. I do believe that there are real benefits to elements of AI in some of these fields.
-
we should consider whether we truly need AI before continuing its development
oops ....
-