IMPORTANT UPDATE ON OPENAI If the last weeks news on OpenAi were confusing and strange to you, you are not alone. Now, we finally have vital piece of information that can help us understand what actually happened and why. Here are the most important insights from the source that was involved in every step of the story. It all stared with the new beta presentation of ChatGPT Med model. This is the first model that actually connects digital and physical world by doing advanced health diagnostics. Basically, it offers something similar to what Tezos was offering: simple test and rich output that could potentially revolutionize health diagnostics and screening, including early detection of some of the most harmful conditions. Because the goal was so similar to notorious Tezos scam, board was skeptical and demanded more information on details. How is it possible that model can detect this? What data is this trained on? Where there any assessment on the error rate, etc. Sam ignored or tried to avoid directly answering these questions and said that everything will be revealed at presentation for the board and investors. All this was happening in the first week of November. The presentation would supposedly show effectiveness of the model as well as all other details. Fast forward to presentation. The model works with urine test and this was one of the scarce pieces of information known to people at the presentation. So, Sam’s assistant took the example urine sample and put it in the machine while Sam typed input: ”Check my overall health”. The results were very detailed and impressive: it showed that the person had undetected insulin resistance that had potential to become diabetes, that the levels of magnesium were below the limit and many other things that could be detected with traditional urine analysis. But the really important find was potential skin cancer and the small suspicious mole was confirmed by further checks. It was not clear that this mole was indeed cancer but the existence of such change on the skin was remarkable achievement. Here is where things become increasingly crazy. Right after that unnamed member of the board stood up and provided sterile cup with the urine sample. What nobody (or almost nobody) knew at that time, was that this person asked his wife, daughter and son for sample and even included urine from his dog and cat. He gave the sample to assistant and typed the prompt himself. Here is the conversation between him and ChatGPT Med A: Why my elbow hurts? ChatGPT Med: Your wife has gonorrhea Your daughter is pregnant Your son is on meth Your dog has fleas Your cat is perfectly healthy. And you might want to be more careful, because you keep hitting your elbow on the bathtub when you frantically masturbate.