FT: Meta develops a rival to ChatGPT

Meta plans to release an AI-powered chatbot capable of supporting multiple personas. This is according to Financial Times.

According to the publication’s sources, the prototype is already in development. Meta’s neural network is expected to debut in September 2023.

The company says the chatbot will be able to converse with users on a human level. The AI’s various archetypes are called ‘personas’ and imitate real or fictional people.

The company has already tested a bot that impersonates the 16th president of the United States, Abraham Lincoln, and has also tested a surfer persona as a travel guide.

According to the FT, the main goal of the product is to improve the recommendation system and search features. The company is automating the checking of the neural network’s responses to ensure accuracy and compliance with rules.

Some experts expressed concerns about the new development. Beyond boosting engagement, chatbots could collect large amounts of data on users’ interests, said Ravit Dotan, AI ethics advisor at the University of Pittsburgh and co-founder of Collaborative AI.

She added that the situation threatens privacy and carries risks of potential ‘manipulation and nudging to action’.

As reported, in April–June 2023 the operating loss of Reality Labs, the unit tied to metaverse within Meta, stood at $3.74 billion.

Earlier, Bloomberg learned about Apple’s plans to create a competitor to ChatGPT. The company has developed a framework for creating large language models called Ajax.