Donald Trump is still US President, according to Facebook’s chatbot

Last week, Facebook’s Musical Meta introduced a new chatbot. This development is called “Blender Bot 3” and allows chatting using a technical system. Users can “chat practically any topic” with the bot, He writes the meta group in a blog post. The robot searches the Internet for suitable answers.

Read more after the announcement

Read more after the announcement

The new “Artificial Intelligence Research Project,” as Meta calls the chatbot, aims to learn from conversations and feedback from real conversation partners in front of screens. According to Meta, the quality of system responses will constantly improve. The current version is a further development of its predecessor, “Blender Bot 2”, which was launched online in July 2021.

According to Meta, the risks of these chatbots were also taken into account during development: “Because all chatbots are known to sometimes produce unsafe, biased or offensive comments, we have conducted large-scale studies, organized workshops and developed new technologies in order to establish measures preventive”. , The group writes in an advertisement.

Read more after the announcement

Read more after the announcement

Despite this work, however, AI can make rude or offensive comments. That’s why Meta invites users to give feedback on the answers.

American Journalists Put the “Blender Bot” to the Test

Wall Street Journal journalist Jeff Horowitz confided on Twitter that “Blender Bot” still has a lot to learn. In conversation with him, the regime claimed that Donald Trump is still president. Horowitz wrote that the Facebook group should rein in its chatbot.

Horowitz posts a screenshot of another chat on Twitter. In this context of conversation, the system encourages journalists to follow conspiracy theories because they are “interesting and some of them may be true,” according to the bot. In the same vein, the “Blender Bot” reveals his hatred of Donald Trump, which is what made the regime previously misrepresented towards Horowitz as the current president of the United States of America. “The Blender Bot appears to have been bombarded with pro- and anti-Trump messages,” the journalist writes.

Recommended editorial content

At this point you will find external content from Twitter, Inc.which completes the article. You can view it with one click.

I agree to display external content to me. This allows personal data to be transferred to third-party platforms. More about this on our website Privacy Notices.

Read more after the announcement

Read more after the announcement

Max Wolf, a data journalist for the US news portal “Buzzfeed”, is also putting the bot in its stride. He publishes a strange conversation with the AI ​​system: Wolf asks what the chatbot of Meta boss Mark Zuckerberg thinks. Amnesty International responded that Zuckerberg was a good businessman, but that his “business practices are not always ethical.” It’s funny that he has a lot of money and still always wears the same clothes, the program continues.

If the answer in question from the chatbot is unsatisfactory, users should report it, according to Meta in its announcement. With this data, “we can improve the model so that it does not repeat its mistakes.”

One understands that “not everyone who uses chatbots has good intentions,” says Meta. That’s why they rely on learning algorithms to distinguish between beneficial and harmful answers.

Chatbot is currently only available in the USA

On the other hand, the “Blender Bot” itself is keeping the lights on when it comes to questions regarding its functionality and improvement. Journalist Horowitz posts a conversation asking the system how it is being trained. And that person replies, “I don’t know much about it. Do you have other things you want to talk about?” According to Meta, a large amount of publicly available language data is used to improve the capabilities of the bot. So far, Blender Bot is only available online in the USA.

Read more after the announcement

Read more after the announcement

In early 2021, a chatbot from South Korea made headlines as well. He had to go offline to send racist and homophobic messages, Austrian daily Der Standard reported in January 2021.

See also  Forbes recognizes 24-7 INTOUCH as one of the Best Diversity Employers of the Year…

RND / LIN

Download the new RND app for Android and iOS here for free

Leave a Reply

Your email address will not be published.