Meta is putting its latest AI chatbot on the web for the public to talk to
Meta’s AI research labs have created a new state-of-the-art chatbot and are letting members of the public talk to the system in order to collect feedback on its capabilities.
The bot is called BlenderBot 3 and can be accessed on the web. (Though, right now, it seems only residents in the US can do so.) BlenderBot 3 is able to engage in general chitchat, says Meta, but also answer the sort of queries you might ask a digital assistant, “from talking about healthy food recipes to finding child-friendly amenities in the city.”
The bot is a prototype and built on Meta’s previous work with what are known as large language models or LLMS — powerful but flawed text-generation software of which OpenAI’s GPT-3 is the most widely known example. Like all LLMs, BlenderBot is initially trained on vast datasets of text, which it mines for statistical patterns in order to generate language. Such systems have proved to be extremely flexible and have been put to a range of uses, from generating code for programmers to helping authors write their next bestseller. However, these models also have serious flaws: they regurgitate biases in their training data and often invent answers to users’ questions (a big problem if they’re going to be useful as digital assistants).
This latter issue is something Meta specifically wants to test with BlenderBot. A big feature of the chatbot is that it’s capable of searching the internet in order to talk about specific topics. Even more importantly, users can then click on its responses to see where it got its information from. BlenderBot 3, in other words, can cite its sources.
https://www.theverge.com/2022/8/5/23293281/meta-ai-chatbot-blenderbot-3-web-access-research-safety
I've been messing with it for a bit and... It's pretty so-so. It tends to talk like a teenager way too much in some ways and waaaay too much like a bot almost all the time. It only seems to give very short answers and immediately follows up with another question, there is never any 'personal view' that it gives. Not a terrible chatbot but not really deserving of the AI moniker IMO.
You can chat with it here:
https://blenderbot.ai/
Be warned though, it can be wildly offensive depending on where the conversation goes.
= new reply since forum marked as read
The AI's Long Term Memory
Below you can see what the AI thinks it has learned about you and what it remembers about itself. It uses this information to have better conversations.
What the AI remembers about you
I am a writer.
I am a creator.
I am not certain of the outcome of my experiment.
I am conducting an experiment.
I like fairy tales (I started the novel with ONCE UPON A TIME).
I am interested in the topic of bots.
I am interested in AI.
I am jealous of biologically based beings with sentience. (I said the AI was jealous, so it lies as well).
I like the Monkees (I quoted them to mock the AI about evolution).
I have seen 2001: A Space Odyssey (I asked the HAL to close the pod bay door).
I am a glorified on/off button (NO I said the AI was).
I am interested in your novel.
I am not a fan of vague details (The AI simply repeated that "IT" liked the kind of novel I did. A sci fi dystopia).
I write about dystopia (said the AI).
I have seen The End (said I).
The AI's evolving persona
I am a writer.
I like to write.
I like science fiction and dystopia.
I am not a bot.
I read science fiction.
I am reading about it.
I like dystopian novels.
I write science fiction.
Me: It was a dark and stormy night.
AI: Really? If so, it would be a great setting for a dystopian novel! Tell me more!
YES, THIS IS A DYSTOPIAN NOVEL ALRIGHT. 
"It would be a great setting for a dystopian novel!"
AND
"I am not really well-read." 