Investments in artificial intelligence systems, which are thought to have an important place in our lives in the future, continue without slowing down. Many technology giants develop their own artificial intelligence and make it available. Chatbots that can communicate with the user via text or speech are also among these systems. The LaMDA model, which has come to the fore with the claims that Google has gained awareness in the past months, can be an example of this. The company denied this and fired the employee who made the claim.
Now, a move has come from Meta, one of the companies that invest in artificial intelligence. The company’s artificial intelligence research lab has developed a state-of-the-art chatbot and made it available online to talk to people. It was stated that the bot was named BlenderBot 3.
Conversations with people will be used to gather feedback on the bot
According to the explanations, BlenderBot 3 is available via this link. However, it should be noted that this access is currently only available to people in the USA, and it may come to other countries in the near future. Meta states that natural and general conversations can be made with the chat bot, and that the system can also be used as a digital assistant. It is stated that the reason for making the bot, which is a prototype, accessible is to gather feedback and provide improvement.
It is also stated that the bot can search the internet to talk about certain topics. It is stated that users can also see where the bot gets the information from. Meta plans to make BlenderBot public, curbing the problems faced by language models. It is reported that the company is trying to minimize responses such as rude language, insults, inappropriate comments, and chat users can flag suspicious replies. Meta officials also say they plan to publicly release all the data they collect during this development phase.
Meta says it won’t experience the chatbot disaster Microsoft experienced years ago
It’s worth noting that rolling out prototype AI bots to the public has had bad consequences for some tech companies in the past. Microsoft recently launched a bot called Tay on Twitter. However, as you can imagine, malicious users made the bot use racist, inappropriate and misogynistic expressions. Microsoft had also taken the bot offline before even 24 hours.
Meta says great advances have been made in artificial intelligence since what happened in Tay. The company states that Microsoft will not repeat this mistake and that BlenderBot will have all security measures to prevent such a thing. Mary Williamson, one of Meta’s artificial intelligence researchers, says Tay learned from real-time communication, while BlenderBot is a static model.
Aimed at bots that can have natural conversations like humans
Williamson adds that most bots used today are task-oriented. Giving an example of customer service bots with pre-programmed dialogs, the researcher emphasizes that what is needed is bots that can chat naturally like humans, noting that this is the purpose of Meta’s work. So what do you think about artificial intelligence chatbots? You can share your thoughts in the comments.
2741
