GPT for ideation

GPT-4 is is a multimodal large language model created by OpenAI and the fourth in its GPT series. It was relased in March this year only a few months after the previous version reflecting the speed at which AI technology is developing. GPT-4 can accept a prompt of text and images, which—parallel to the text-only setting—lets the user specify any vision or language task. Like previous GPT models, the GPT-4 base model was trained to predict the next word in a document, and was trained using publicly available data (such as internet data) as well as licensed data. Although less capable than humans in many real-world scenarios, the model exhibits human-level performance on various professional and academic benchmarks, for example, simulating exams that were originally designed for humans resulted in a 90 per cent score in law exams, and art history. Despite its capabilities, however, GPT-4 has similar limitations as earlier GPT models. Most importantly, it still is not fully reliable, for example, it “hallucinates” facts and makes reasoning errors. Also, the model can have various biases in its outputs. So when prompted with a question, the base model can respond in a wide variety of ways that might be far from a user’s intent. Such responses may, or may not trigger new ideas. The language model, however, points to creative futures when AI will help turn designing into a seamless process from idea to realisation, from first thought to production and manufacturing. Indeed, the design process, including ideation is already largely driven by language (verbal and visual).