OpenAI Withholds Powerful Text Algorithm Over Safety Fears
OpenAI reveals a highly capable text-generating algorithm called GPT-2 but refuses to release the full model due to safety concerns. The decision sparks debate among experts about whether the threat is exaggerated and how to handle dangerous AI.
OpenAI unveils a new text-generation model called GPT-2 that writes coherent and versatile prose based on simple prompts. Citing safety and security concerns, the nonprofit research group decides to withhold the full algorithm, the training data, and the code from the public.
Media coverage quickly paints a frightening picture of the technology, with headlines warning of a locked-up AI that threatens humanity and suggesting a robot apocalypse. However, machine learning experts actively debate whether these alarming claims are actually exaggerated.
This strategic withholding by OpenAI, an organization founded by Elon Musk and other tech leaders to guide responsible AI development, sparks a broader industry conversation. The tech community now faces difficult questions about how to manage the proliferation of potentially dangerous artificial intelligence algorithms.