OpenAI said it is developing a new tool that can determine with a high degree of accuracy whether a picture was drawn by artificial intelligence.
In addition to the chatbot and writing tool ChatGPT, OpenAI has also developed an artificial intelligence image generator called “DALL-E”.
Image source Pexels
Mira Murati, the company’s chief technology officer, revealed that the AI picture recognition tool developed by the company can determine with 99% accuracy whether a picture was drawn by an artificial intelligence tool.
Mira Mulati said that this image recognition tool is undergoing internal testing and will be released later, but she did not provide a very specific release time.
OpenAI’s ChatGPT has brought a revolution in generative artificial intelligence technology. Many products on the market can generate images based on keywords entered by users. On the other hand, some artificial intelligence companies have also launched recognition tools to determine certain images or articles. Whether the content was created by artificial intelligence tools, but these tools are not very accurate.
At the beginning of this year, OpenAI also launched an identification tool, which mainly identifies whether an article was created by an artificial intelligence assistant. However, unexpectedly, this product was withdrawn from the list in July because the accuracy was not high and it was not suitable for users. Not reliable yet.
When it was withdrawn, OpenAI stated that it would continue to improve this recognition product in the future, and also recognize content created by other artificial intelligence assistants, such as pictures, audio, etc.
Although generative artificial intelligence brings convenience to many people, it also brings other worries. For example, in the field of journalism, when major events occur, some bad actors will use artificial intelligence tools to fabricate news articles and spread them around. This creates a need for identification tools.
The creation of artificial intelligence also brings another problem, which is that it damages the copyright of other works. Recently, the Firefly image generator team owned by Adobe in the United States announced that it will never allow images created by artificial intelligence to infringe the copyrights of others.
OpenAI’s large language model and ChatGPT are industry-star products. On Tuesday, the company’s executives also talked about subsequent version plans for the GPT-4 large language model.
So far, OpenAI has not announced the name of the subsequent large language model, but data shows that in July this year, the company applied for a new trademark – “GPT-5” with the U.S. Patent and Trademark Office.
As many users have discovered, ChatGPT, the artificial intelligence assistant based on GPT-4 and GPT-3.5 large language models, will give wrong answers and fall into the so-called “artificial intelligence illusion.” Regarding the media’s question on whether the GPT-5 large language model will continue to give wrong answers, Mulati’s answer was “possibly.”
Mulati said that the development team has made some progress on the problem of artificial intelligence hallucinations in the GPT-4 model, but it is clear that the company is not yet where it should be.
It is reported that artificial intelligence manufacturers need to use high-performance processors to train large language models, and in this processor market, Nvidia is currently the leader. On Tuesday, Sam Altman also talked about whether OpenAI will develop the processor chips needed for training itself.
Sam Altman said: “The default strategy is that we will definitely not develop chips, but I am not willing to rule it out.”
THIS IS A SPONSOR PROMOTION: >>>>>>>>>>>>>
Geekwills is an online shop that connects consumers with millions of products and brands around the world with the mission to empower them to live their best lives. Geekwills is committed to offering the most affordable quality products to enable consumers and sellers to fulfill their dreams in an inclusive environment.