ChatGPT users have become frustrated with the AI model's tone, and OpenAI is taking action. After widespread mockery of the robot's relentlessly positive and complimentary output recently, OpenAI CEO Sam Altman confirms the company will roll back the latest update to GPT-4o. So get ready for a more reserved and less sycophantic chatbot, at least for now.

GPT-4o is not a new model—OpenAI released it almost a year ago, and it remains the default when you access ChatGPT, but the company occasionally releases revised versions of existing models. As people interact with the chatbot, OpenAI gathers data on the responses people like more. Then, engineers revise the production model using a technique called reinforcement learning from human feedback (RLHF).

Recently, however, that reinforcement learning went off the rails. The AI went from generally positive to the world's biggest suck-up. Users could present ChatGPT with completely terrible ideas or misguided claims, and it might respond, "Wow, you're a genius," and "This is on a whole different level."

Read full article

Comments

ChatGPT users have become frustrated with the AI model's tone, and OpenAI is taking action. After widespread mockery of the robot's relentlessly positive and complimentary output recently, OpenAI CEO Sam Altman confirms the company will roll back the latest update to GPT-4o. So get ready for a more reserved and less sycophantic chatbot, at least for now.

GPT-4o is not a new model—OpenAI released it almost a year ago, and it remains the default when you access ChatGPT, but the company occasionally releases revised versions of existing models. As people interact with the chatbot, OpenAI gathers data on the responses people like more. Then, engineers revise the production model using a technique called reinforcement learning from human feedback (RLHF).

Recently, however, that reinforcement learning went off the rails. The AI went from generally positive to the world's biggest suck-up. Users could present ChatGPT with completely terrible ideas or misguided claims, and it might respond, "Wow, you're a genius," and "This is on a whole different level."

Read full article

Comments

ChatGPT users have become frustrated with the AI model's tone, and OpenAI is taking action. After widespread mockery of the robot's relentlessly positive and complimentary output recently, OpenAI CEO Sam Altman confirms the company will roll back the latest update to GPT-4o. So get ready for a more reserved and less sycophantic chatbot, at least for now.

GPT-4o is not a new model—OpenAI released it almost a year ago, and it remains the default when you access ChatGPT, but the company occasionally releases revised versions of existing models. As people interact with the chatbot, OpenAI gathers data on the responses people like more. Then, engineers revise the production model using a technique called reinforcement learning from human feedback (RLHF).

Recently, however, that reinforcement learning went off the rails. The AI went from generally positive to the world's biggest suck-up. Users could present ChatGPT with completely terrible ideas or misguided claims, and it might respond, "Wow, you're a genius," and "This is on a whole different level."

Read full article

Comments