Meta’s AI research labs have reportedly created a new state-of-the-art chatbot, and the tech giant is letting members of the public talk to the system to get feedback on its capabilities.
According to The Verge, the bot is called BlenderBot 3 and is accessible on the web. Although right now it looks like only US residents can do it.
“We’re committed to publicly releasing all the data we collect in the demo in the hope that we can improve conversational AI,” said Kurt Shuster, a research engineer at Meta who helped build BlenderBot 3.
BlenderBot 3 is able to engage in general chit-chat, Meta said, but also answer questions you might ask the digital assistant, “from talking about healthy food recipes to finding kid-friendly facilities in town,” according to the report.
The bot is a prototype and builds on Meta’s previous work with what are known as large language models, or LLMS – powerful but flawed text generation software, the most famous example of which is OpenAI GPT-3.
Like all LLMs, BlenderBot is initially trained on large text datasets, from which it benefits from statistical patterns to generate language.
Such systems have proven to be extremely flexible and have found a variety of uses, from generating code for programmers to helping authors write their next bestseller.
However, these models also have serious flaws: they reject biases in their training data and often invent answers to user questions (a big problem if they are to be useful as digital assistants).