OpenAI CEO expresses concern over potential misuse of GPT-4 AI technology.
Researchers reflect on their 2021 paper "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?" which warned of the risks associated with large language models (LLMs) such as GPT-4. The paper argued that much of the text mined to build GPT-3 comes from forums that do not include the voices of women, older people, and marginalized groups, leading to inevitable biases. OpenAI CEO Sam Altman recently admitted the company is "a little bit scared" of GPT-4's risks, including the "hallucinations problem." The researchers recommend mitigating risks by carefully curating and documenting datasets and encouraging research beyond ever-larger language models.