Suara Malaysia
ADVERTISEMENTFly London from Kuala LumpurFly London from Kuala Lumpur
Monday, December 23, 2024
More
    ADVERTISEMENTFly London from Kuala LumpurFly London from Kuala Lumpur
    HomeTechOpenAI claims tool to detect AI-generated images is 99% accurate

    OpenAI claims tool to detect AI-generated images is 99% accurate

    -

    Fly AirAsia from Kuala Lumpur

    OpenAI is developing a highly accurate tool to detect images created using artificial intelligence (AI). Mira Murati, the chief technology officer behind popular AI chatbot ChatGPT and image generator Dall-E, announced on Tuesday (Oct 17) that OpenAI’s tool is “99% reliable” in identifying AI-generated pictures. The tool is currently undergoing internal testing and will be released to the public in the future, although no specific timeline was provided.

    Murati made the announcement during the Wall Street Journal’s Tech Live conference in Laguna Beach, California, where she was joined by OpenAI CEO Sam Altman. Several existing tools claim to have the ability to detect AI-generated content, but they often lack accuracy. OpenAI previously released a similar tool in January to determine AI-generated text but discontinued it in July due to its unreliable performance. However, the company has been working on improving the software and plans to develop methods for identifying AI-generated audio and images.

    With the increasing use of AI in creating and manipulating news reports, the demand for detection tools has become vital. Adobe Inc’s Firefly image generator addresses one aspect of this challenge by promising to create content without infringing on intellectual property rights.

    During the conference, Altman and Murati also hinted at the potential successor to GPT-4, OpenAI’s current AI model. While the company has not publicly announced the name of the follow-up model, they filed a trademark application for “GPT-5” with the US Patent and Trademark Office in July.

    Chatbots like ChatGPT, which utilize GPT-4 and its predecessor GPT-3.5, are susceptible to generating false information, also known as hallucination. When asked about the possibility of GPT-5 eliminating this issue, Murati responded, “Maybe.” She acknowledged significant progress in addressing hallucination problems with GPT-4 but admitted that further improvements are necessary.

    ALSO READ:  Analysis-For retail investors, jumping on Arm's blockbuster IPO is a risky business

    Altman also discussed the potential of OpenAI designing and manufacturing its own computer chips for training and operating AI models, rather than relying on companies like Nvidia Corp, which currently dominates the market.

    He stated, “The default path would certainly be not to, but I would never rule it out.” – Bloomberg

    Suara
    Suarahttps://www.suara.my
    Tech enthusiast turning dreams into reality, one byte at a time 🚀

    Related articles

    ADVERTISEMENTFly London from Kuala Lumpur

    Subscribe to Newsletter

    To be updated with all the latest news, offers and special announcements.

    Latest posts