Communication

The world of AI is fast-moving and constantly evolving, with new developments and technologies emerging every day. This week saw multiple stories that caught the attention of industry experts and observers alike. One such story involved ChatBots multiplying at an alarming rate, with new models and iterations being released regularly. However, not all of these models are equal in accuracy or reliability, as evidenced by the recent report on ChatGPT’s inaccuracies in Chinese dialects compared to English. Another example of the technical flaws that are still present in generative AI can be seen in Hugging Face’s conversational AI, HuggingChat, which has been found to give inconsistent and even bizarre answers to basic questions. Despite these challenges, there are renewed efforts to improve text-generating AI, such as Nvidia’s NeMo Guardrails, which aims to make AI safer and more reliable through open-source code and documentation. As the quest for maximum truth-seeking in AI continues, it is important to remain vigilant and aware of the limitations and biases that still exist in these systems.