On the other hand, this feature will allow the chatbot to have the ability to maintain longer conversations with users, as it will be able to remember what they commented at the beginning of a conversation, for example. The integration of GPT in Bing, to mention one case, is limited to remembering what he said in a span of 20 turns.
It is worth mentioning that this innovation has a limitation and that is that it is not available to all Anthropic users, but only to the company’s business partners, who would also have to spend a greater amount of money to obtain the changes.
Misinformation, a persistent problem
Although this represents an advance for AI models, it still has a problem to solve and that is the tendency of chatbots to make up information. Given this, the company announced a “constitution” to generate a safe and reliable AI.
Anthropic’s research in this regard proposes that the AI provide safer and more useful answers, since during its training the model received input from a series of sources with ethical content regarding security and privacy.
