Here’s how OpenAI plans to clean up ChatGPT from misinformation

On 31 May, OpenAI announced its efforts to enhance the mathematical problem solving capabilities of ChatGPT with the aim of reducing instances of artificial intelligence (AI) hallucinations. OpenAI emphasizes reducing hallucinations as an important step towards the development of aligned AI.

In March, the introduction of the latest version of ChatGPT – ChatGPT-4 pushed AI further into the mainstream. However, generative AI chatbots have long struggled with factual accuracy, sometimes generating false information, commonly referred to as “hallucinations”. There were attempts to reduce these AI hallucinations announced Via a post on OpenAI’s website.

AI hallucinations refer to instances where artificial intelligence systems generate factually incorrect outputs, misleading or unsupported by real-world data. These hallucinations can appear in various forms, such as generating false information, creating non-existent events or people, or providing false descriptions about certain subjects.

OpenAI operated Research to test the effectiveness of two types of feedback: “outcome supervision” and “process supervision.” Outcome supervision involves feedback based on the end result, while process supervision provides input for each step in the chain of thought. OpenAI evaluated these models using math problems, generated multiple solutions, and selected the highest ranked solution according to each feedback model.

After a thorough analysis, the research team found that process supervision delivered better performance because it encouraged the model to follow a human-approved process. In contrast, outcome supervision proved more challenging to examine consistently.

OpenAI recognizes that process supervision has implications beyond mathematics, with further investigation necessary to understand its implications in different domains. This raised the possibility that process supervision may offer a favorable combination of performance and alignment compared to outcome supervision if the observed results are in broader contexts. To facilitate research, the company publicly released the complete data set of process supervision, inviting exploration and study in this area.

Connected: AI demand briefly propels Nvidia into the $1T club

Although OpenAI did not provide clear examples of what inspired its investigation into hallucinations, two recent incidents exemplified the problem in real-life scenarios.

In a recent incident, attorney Steven Schwartz in the case Mata v. Avianca Airlines accepted Relying on chatbots as a research resource. However, the information provided by ChatGPT turned out to be completely fabricated, thereby highlighting the issue.

OpenAI’s ChatGPT isn’t the only example of an artificial intelligence system encountering hallucinations. during a Display In March in its chatbot technology, Microsoft’s Bing AI chatbot probed earnings reports and generated false figures for companies like Gap and Lululemon.

magazine: 25K traders bet on ChatGPT’s stock picks, AI sucks at rolling the dice, and more