OpenAI introduces GPT-4 – how long ago was it?
The background
On March 14, 2023, OpenAI introduced the GPT-4 AI model. It could process sophisticated texts and, in an additional feature presented, could also evaluate images as input. However, this image function was initially not generally available. The new model became gradually accessible via ChatGPT Plus and a programming interface. The publication increased the competition for generative artificial intelligence, i.e. systems that generate text or other content themselves. For companies, schools and authorities in Germany and around the world, questions about productivity, data protection and reliable information became more urgent. Despite better performance, GPT-4 remained error-prone: convincingly formulated answers could still be factually incorrect. An advance in the language model was therefore no guarantee of truth or human understanding.