I like Claude, Anthropic’s generative algorithm

Over the last few weeks I’ve been trying out Claude, the generative algorithm launched by Anthropic in March and now in its second iteration, and in my opinion this is one of the most interesting players in an increasingly crowded field, with Bard, that Google has recently introduced in 43 languages and in many more countries, and Meta’s LLaMA, which is about to be launched commercially for corporate customers.

What makes Claude interesting? First of all, the company behind it. Anthropic is a public benefit corporation based in San Francisco focused on developing reliable, interpretable and steerable AI systems, responsible and secure treatment of data, based on its three Hs model: helpful, honest and harmless. These principles, far from the unbridled ambition of OpenAI or other companies, have allowed the company to raise $1.5 billion; its last fundraisinground garnered $450 million in May. Furthermore, it is committed to transparency in the characteristics of its model card, as well as in its API and, also important, its price.

Anthropic has carried out the development of Constitutional AI, a model that seeks to align artificial intelligence systems with universal human values and promote a positive social impact, subject to the following principles:

• Alignment with the UN Universal Declaration of Human Rights…