A team of Microsoft researchers has presented a new AI that is even more versatile than the famous ChatGPT. She is not only able to answer text questions, but she can also analyze and process images.
In an article titled ” Language Isn’t All You Need: Aligning Perception and Language Patterns (Language Is Not All You Need: Aligning Perception with Language Models), Microsoft researchers presented Kosmos-1. As they recall in the introduction to their paper, “a great convergence of language, multimodal perception, action and modeling of the world is a key step towards generalist artificial intelligence”.
To read – ChatGPT puts more and more employees out of work
As we know, Microsoft appreciates ChatGPT so much that it has invested several billion dollars in OpenAI, the designer of the chatbot. The Redmond firm wishes to integrate this invention into several of its software. The company’s search engine, Bing, uses GPT 3.5, an evolution of the language model used by ChatGPT. But the Redmond firm does not intend to stop there. A true generalist AI will not be able to content itself with understanding a text. She should be able tointerpret images and their contexts, or to understand sound signals. In other words, she will have to use the five senses just like a human being.
Kosmos-1 will succeed ChatGPT, it includes texts and images
There new AI presented by Microsoft, is able to do the same thing as ChatGPT, and more. Indeed, Kosmos-1 is not just a conversational agent. According to its creators, it achieves impressive performance in language comprehension and generation, multimodal dialogue, image captioning, answering visual questions, and vision tasks (such as classification according to text instructions)”.
We are therefore getting even closer to the famous concept of generalist artificial intelligence., which can replace the human being in all intellectual tasks. Should we be happy or worried? One thing is certain: AI is already transforming our society.