More than half a decade after Microsoft’s truly monumental Taye debacle, the incident remains a stark reminder of how quickly an AI can become corrupted after exposure to the potent toxicity of the internet, and a warning against building untethered bots. strong enough behaviour. On Friday, Meta’s AI research division will see if its latest version of Blenderbot AI can stand up to the online horrors with the public demo release of its 175 billion parameter Blenderbot 3.
A major hurdle currently facing chatbot technology (as well as the natural language processing algorithms that power them) is sourcing. Traditionally chatbots are trained in highly curated environments, because otherwise you invariably get a Taye, but that ends up limiting the topics you can discuss to those specific ones available in the lab. Rather, you can have the chatbot pull information from the internet to gain access to a wide range of topics, but it could, and probably will, go full Nazi at some point.
“Researchers cannot predict or simulate all conversation scenarios alone in research settings,” Meta AI researchers wrote in a blog post on Friday. “The field of AI is still far from truly intelligent AI systems that can understand us, interact and converse with us like other humans can. To build models that are more adaptable to real-world environments, chatbots need to learn from a wide-ranging and diverse perspective with people ‘in the wild'”.
Meta has been working to address the issue since it first introduced the BlenderBot 1 chat app in 2020. Initially little more than an open source NLP experiment, by the following year BlenderBot 2 had learned to remember information it had discussed in previous conversations. and how to search the Internet for additional details on a given topic. BlenderBot 3 takes those capabilities a step further by evaluating not only the data you pull from the web, but also the people you talk to.
When a user logs an unsatisfactory response from the system, which currently hovers around 0.16 percent of all training responses, Meta processes the user’s feedback into the model to avoid repeating the error. The system also employs the Director algorithm, which first generates a response using training data, then runs the response through a classifier to check if it fits within a scale of correct and incorrect defined by user feedback.
“To generate a sentence, the language’s classification and modeling mechanisms must agree,” the team wrote. “Using data that indicates good and bad responses, we can train the classifier to penalize low-quality, toxic, contradictory, or repetitive statements, and statements that are generally not helpful.” The system also employs a separate user weighting algorithm to detect unreliable or ill-intentioned responses from the human chatter, essentially teaching the system not to trust what that person has to say.
“Our live, interactive, public demo allows BlenderBot 3 to learn from organic interactions with all kinds of people,” the team wrote. “We encourage adults in the United States to try the demo, have natural conversations about topics of interest, and share their responses to help advance the research.”
BB3 is expected to speak more naturally and conversationally than its predecessor, in part, thanks to its vastly updated OPT-175B language model, which is nearly 60 times larger than BB2’s model. “We found that, compared to BlenderBot 2, BlenderBot 3 provides a 31 percent improvement in the overall score for conversation tasks, as assessed by human judgments,” the team said. “He is also considered to be twice as knowledgeable, while the facts are wrong 47% less of the time. Compared to GPT3, he is more up-to-date 82% of the time and more specific 76% of the time on topical questions.” weather.”
All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission.