ChatGPT and its brothers are both surprisingly smart and disappointingly dumb. Sure, they can generate cute poems, solve scientific puzzles, and debug spaghetti code. But we know that they often make, forget and act like weirdos.
tipping AIa company founded by researchers who previously worked on major artificial intelligence projects in Google, open AIand nvidiabuilt a bot called Pi that seems to make fewer mistakes and be more adept at sociable conversations.
Inflection designed Pi to address some of the problems with today’s chatbots. Programs like ChatGPT use artificial neural networks that try to predict which words should follow a piece of text, such as an answer to a user’s question. With enough training in billions of lines of human-written text, backed by high-powered computers, these models can generate coherent, relevant responses that feel like real conversation. but they too make things up and go off the rails.
Mustafa Suleyman, Inflection’s CEO, says the company has carefully curated Pi training data to reduce the chance of toxic language creeping into its responses. “We’re pretty selective about what goes into the model,” he says. “We take a lot of information that’s available on the open web, but not absolutely everything.”
Suleyman, who co-founded the artificial intelligence company Deepmind, which is now part of Google, also says that limiting the length of Pi’s responses reduces, but doesn’t completely eliminate, the likelihood of factual errors.
Based on my own time chatting with Pi, the result is attractive, if more limited and less useful than ChatGPT and Bard. Those chatbots became better at answering questions through additional training in which humans rated the quality of their answers. That feedback is used to guide the bots towards more satisfying answers.
Suleyman says Pi was similarly trained, but with an emphasis on being friendly and supportive, albeit without a human personality, which could confuse users about the program’s capabilities. Chatbots that take on a human persona have already proven problematic. Last year, a Google engineer controversially claimed that the company’s AI model LaMDAone of the first programs to demonstrate how smart and attractive large AI language models could be, it could be sentient.
The Pi can also keep track of all your conversations with a user, giving it a kind of long-term memory that ChatGPT lacks and aims to add consistency to your chats.
“A good conversation is about responding to what a person is saying, asking clarifying questions, being curious, being patient,” says Suleyman. “It’s there to help you think, rather than give you strong direction advice, to help you unpack your thoughts.”
Pi adopts a caring and talkative personality, even if she doesn’t pretend to be human. She would often ask me how she was doing and often offer words of encouragement. Pi’s short responses mean she would also work well as a voice assistant, where long responses and mistakes are especially jarring. She can try to speak it yourself on the Inflection website.
The incredible hype around ChatGPT and similar tools means that many entrepreneurs are hoping to strike it rich in the field.
Suleyman used to be a manager within the Google team working on the LaMDA chatbot. Google was hesitant to release the technology, much to the frustration of some of those working on it who believed it had great commercial potential.
—————————————————-
Source link