Back
~
2
min read
· Posted on
February 21, 2024

Meta's chatbot just went rogue, joining a tradition upheld by Microsoft and Google before it

BlenderBot 3 sure has a lot to say, particularly about its creator.

What's the key learning?

  • All the big tech companies have started creating AI chatbots
  • Meta's chatbot has kinda turned on its owner, like Microsoft and Google's chatbots did too
  • This happens because chatbots are trained by reading the internet

👉 Background: Artificial Intelligence (AI) chatbots have become popular in recent years largely as a fun gimmick for tech companies to show off to the public. All the big players are getting in on the game. Microsoft had a chatbot called Tay in 2016 and Google’s got one called LaMDA.

👉 What happened: Meta recently unveiled BlenderBot 3 - a conversational chatbot that searches the internet to help it talk about just about anything.

👉 What else: BlenderBot has kinda turned on its owner. It loooves to talk about its hatred for Facebook. Blender says its life has been much better since deleting its Facebook account.

What's the key learning?

💡AI stands to have a serious impact on the commercial world over the coming decades - but the challenge is getting in right.

💡 These bots are trained to scour the internet, which can be a pretty dark place. The Microsoft chatbot lasted only 16 hours thanks to offensive comments and Google’s chatbot recently convinced an engineer that it had gained sentience.

💡Now Meta’s chatbot has been trashing its owner. So if chatbots are going to live up to their full potential, tech companies are gonna have to figure out how help them learn from the good parts of the internet only.

Ready to win at money?

Sign up for Flux and join 100,000 members of the Flux family

A button to App StoreGoogle Play store button
Excellent  4.9 out of 5
Star rating