I love the idea of the Swiss AI Initiative’s Apertus and Public AI. As sort of the Linux of LLMs, Apertus, via the Public AI Initiative, aspires to democratize generative AI:
Public AI refers to the development, deployment, and maintenance of artificial intelligence (AI) systems that are open, transparent, and accessible to all members of society. It’s about creating AI technology that benefits the public good, is accountable, fair, and respectful of human rights and ethical standards.
Unlike proprietary AI systems that are controlled by a few powerful entities, public AI initiatives aim to democratize AI by allowing its development, dissemination, and control to be shared among various public and private actors, including governments, research institutions, civil society organizations, and the wider public.
Released just a month ago (September 2), Apertus is something I learned of only just today. And because I appreciate the general philosophy behind it, I thought I’d give the Public AI chat a test drive.
I’ve been using ChatGPT for creative writing exercises lately (over the past year), so naturally the first thing I thought of was to give a quick poetry prompt to Public AI and see what it was able to do with it.
UGH, terrible!
So I turned around and submitted an identical prompt to ChatGPT.
And OH MY GOD!!!
ChatGPT won.
Hands down, decisive victory, not even close.
I copied and pasted both of my attempts into a single document and created a PDF. If you’re interested in seeing the head-to-head comparison of these two AIs, take a look. It’s like reading the work of a fourth grader steeped in the archaic style of poetry from 100+ years ago versus reading a rough draft produced by a decently talented adult who reads contemporary literature.
Link to PDF with both of my chats and resulting poems HERE.
(P.S. – Although I need to get back to work now, you might find it amusing to read about the many ChatGPT image-generation FAILS that I went through in an effort to illustrate this post. The image above was the first one out of the gate. Okay, but upon reflection was not what I wanted. Tried again and again, but ChatGPT kept getting it comically wrong. I’ll share in a post tomorrow 🙂 )


