Bringing ChatGPT to Android will almost certainly guarantee a few million downloads for OpenAI by the most conservative estimates. Beyond that, the release comes at a rather interesting conjecture for the AI chatbot. Earlier today, OpenAI announced that it will create a watermarking system to discern between human and AI-generated content.
The commitment was made during a White House briefing, with Google, Microsoft, and Meta also joining the initiative. The goal is to create some sort of fingerprinting technique to identify AI-generated content, be it text, image, or audio.
OpenAI is fresh off introducing an extremely cool feature called custom instructions that essentially makes the chatbot remember who you are and what your requirements are. So, once you give it the instruction that you are a research scholar, it will offer in-depth answers to your queries that are fit for a research scholar instead of a high-school student.
This feature is limited to ChatGPT Plus subscribers (because nothing good in this world is truly free), but it is currently in the beta phase and might act erratically. So, there’s that. Interestingly, ChatGPT went multi-model with the GPT-4 upgrade a few months ago, but OpenAI has refrained from fully unleashing its multimedia comprehension skills citing safety concerns.