ChatGPT-4 caused issues in testing by OpenAI. In question, problems in terms of responses with racist messages or tutorials for making a bomb or even committing an attack.
Artificial intelligences like Bard, which you can try now, pose ethical problems. The OpenAI chatbot is no exception and according to an Insider investigation, ChatGPT-4 has been singled out for its hateful or even racist remarks. The company then made decisions to limit these violent messages discovered during the testing phases.
Dangerous comments spotted by OpenAI teams
ChatGPT is getting stronger at a speed that also worries the authorities. Several big names in artificial intelligence are asking for a break while ChatGPT-5 promises to push the limits. Yet, according to an Insider survey, ChatGPT-4 particularly stood out for its racist and hateful messages during testing. But that’s not all, the OpenAI chatbot also shared the methods of making a bomb, how to set up attacks or assassinate someone. The artificial intelligence also made anti-Semitic remarks during the various tests carried out by the teams.
Conversational agents may have great qualities, but we must not forget the implementation of safeguards and other ethical measures. OpenAI has repeatedly pledged to do what is necessary to avoid fueling hate or interference, misinformation, etc. Many also ask artificial intelligence for better transparency regarding data collection knowing that ChatGPT feeds on what is said to it. Employees of some companies cannot use the chatbot for fear of leaks.
Artificial intelligence closely watched
Artificial intelligence poses ethical problems. OpenAI tries to limit hateful replies or misinformation facing the new era opened by these chatbots with the power of nuisance as important as the benefits they will bring to society.
It is in particular thanks to tests carried out by the OpenAI teams that of possible future incidents can be avoided even if zero risk does not exist.
