The fight for information has become more fierce than ever. The rise of artificial intelligence has led big tech companies to seek all kinds of ways to gather more data and, in this way, train their models. However, Meta knows that it has an advantage in this game, thanks to Instagram and Facebook, and wants to use it to its advantage to train Emu, its image generator.
Chris Cox, product chief at Meta, stated during the Bloomberg Tech Summit, held last Thursday, that the data used to train Emu is completely public. According to what we can read on Business Insider, Cox explained that they do not use private information like what we can “share with our friends,” for example.
According to Cox, the model being developed by Meta can produce “images with incredible quality” thanks to the huge amount of photos uploaded to Instagram. After all, the social network has served for many years as a showcase for the creative sector, whether they are photographers, designers, artists, and so on. Now, Meta wants to repay the favor with a very high interest.
Meta AI, Meta’s ChatGPT
Fortunately (or unfortunately), you won’t have to wait for Meta to launch Emu to generate images: you can do it from Meta AI itself by simply typing the word “imagine“. Then, you just have to describe what you want to obtain and the bot will provide four results. However, Meta AI can only be used if we are in the United States, Australia, Canada, Ghana, Jamaica, Malawi, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Uganda, Zambia or Zimbabwe.
Meta AI works thanks to LlaMA 3, the latest version of Meta’s artificial intelligence system and direct competitor to GPT-4 and Google Gemini. Gradually, we will see how Meta AI will reach Facebook, WhatsApp, and Instagram globally as if it were a conversational bot to chat with.
Source: Business Insider
