Open AI has released its latest AI model GPT-4, a large multimodal model that can assimilate both image and text inputs to create text outputs. The AI is touted to display human-level performance on many academic and professional benchmarks and outperforms all existing large language models. The software is an improved version of GPT-3.5 regarding reliability, creativity, and dexterity in instruction handling. It is also said to be much more secure than its predecessor and can handle many languages, including low-resource ones like Latvian, Welsh, and Swahili. The AI is highly steerable, allowing developers to set its style and task by detailing the directions in the “system” message. On the downside, the software has no knowledge of incidents after September 2021, thereby running the risks of simple reasoning errors and hallucination of facts.
Open AI Releases Next-level AI GPT-4 Available Through ChatGPT and Bing
March 15, 2023