A few months ago, major tech company OpenAI launched ChatGPT and it quickly became a viral chatbot tool. Since then, it has impressed everyone by creating original essays, short stories, instruction sets and even coding. Users can run it for free as long as they create a personal account. They can simply type their request, and ChatGPT will execute it for them.
On Tuesday March 28, OpenAI announced the release of GPT-4, a newer version with advanced capabilities. According to the developers who designed it, GPT-4 can perform well on standardized tests. It actually passed a simulated law school bar exam with a score around the top 10%. The previous version, GPT-3, only made it to the bottom 10% of the scores when trying to pass the very same test. However, some people are concerned about the risks behind ChatGPT. For instance, it can go off the guardrails with its responses sometimes. Now and then, it makes factual errors, reacts "emotionally" and engages in "hallucinations".
OpenAI said GPT-4 is already available via wait list. This version "hallucinates" less and builds more creative models. It is also making its way into some third-party products like Microsoft’s AI-powered Bing.
