Claude and "Constitutional" AI

For a while now, I have been of the firm opinion that we need to build in Asimov's Three Laws of Robotics into our AI models, especially given the speed of advancement of LLMs over the past couple of years. Am glad that folks are finally starting to do that.

Anthropic, a Google-backed AI company started by ex-OpenAI folks has just announced the release of Claude, a rival to chatGPT.

So how is Claude different? From the article:

Anthropic says that Claude — which, like ChatGPT, doesn’t have access to the internet and was trained on public webpages up to spring 2021 — was “trained to avoid sexist, racist and toxic outputs” as well as “to avoid helping a human engage in illegal or unethical activities.” That’s par for the course in the AI chatbot realm. But what sets Claude apart is a technique called “constitutional AI,” Anthropic asserts.

“Constitutional AI” aims to provide a “principle-based” approach to aligning AI systems with human intentions, letting AI similar to ChatGPT respond to questions using a simple set of principles as a guide. To build Claude, Anthropic started with a list of around 10 principles that, taken together, formed a sort of “constitution” (hence the name “constitutional AI”). The principles haven’t been made public. But Anthropic says they’re grounded in the concepts of beneficence (maximizing positive impact), nonmaleficence (avoiding giving harmful advice) and autonomy (respecting freedom of choice).

So looks like folks are finally trying to put in some version of Asimov's Three Laws of Robotics:

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey orders given it by human beings except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

The concepts of "beneficence seems to line up with Second and Third Laws. Nonmaleficence is clearly the First Law. Autonomy again lines up with the Third Law.

These are still early days, and, as the article states, there are ways to engineer prompts to get around the "Constitutional" limitations imposed by Claude, but perhaps this is a harbinger of how we need to proceed. Unfettered AI can have significant negative impacts on our societies. We must ensure, for our future generations that we continue to develop these technologies with built-in safety mechanisms.

Claude seems to be a good start.