OpenAI is expected to generate over $1 billion in annual revenue, surpassing previous expectations. The company, known for its AI chatbot ChatGPT, currently earns around $80 million monthly. Last year, OpenAI’s revenue was $28 million before it began charging for ChatGPT. The chatbot has gained popularity due to its ability to mimic human conversation and perform various tasks. However, some users have found the bot “unhinged” and argumentative.
The success of ChatGPT has attracted significant investments in AI. OpenAI recently released a business version of ChatGPT, called ChatGPT Enterprise, catering to Fortune 500 companies. The company has also introduced a premium subscription service for dedicated users. OpenAI generates revenue from ChatGPT and by selling API access to its AI models. However, a deal with Microsoft entitles the tech giant to 75% of OpenAI’s profits until its $13 billion investment is repaid. OpenAI’s CEO, Sam Altman, expects the company’s ambitions around artificial general intelligence (AGI) to cost another $100 billion.
Director Gareth Edwards discusses the release timing of his film, The Creator, which explores the war between humans and AI and draws parallels to the strikes by SAG-AFTRA and WGA. Edwards jokes about the film’s timing, saying he aimed for a “sweet spot” window before the apocalypse, around November or December. He mentioned that recent events have made the film’s premise feel even scarier and weirder, and actors and writers have been on strike seeking protection from AI.
Edwards revealed that the movie’s setup reflects the last few months, addressing executives’ inquiries about the backstory behind the war between humans and AI. He described The Creator as a mix of Blade Runner and Apocalypse Now, influenced by the works of Francis Ford Coppola and Ridley Scott. Edwards hopes the film will evoke empathy for others and shed light on the fascinating consequences of creating AI. He also mentioned the casting of Madeleine Yuna Voyles in the role of Alfie.
AI is causing a shift in the web’s dynamics as big tech companies freely utilize content without compensating creators. Web crawlers like Common Crawl and GPTbot scrape vast amounts of data from the internet, which is then used for training AI models. This undermines the original purpose of web crawlers, which was to index information for search engine results. Content creators are starting to block these crawlers using a simple code called robots.txt, but it is an imperfect and easily bypassed tool.
The lack of enforceability means that crawlers can still collect data without consent. Major companies have started using robots.txt to block GPTbot and Common Crawl, but the issue remains unresolved. Creative Commons, an organization that licenses content for use on the internet, is exploring new strategies to address the problematic use of open content for AI training. There are concerns that if the situation is not rectified, the internet could become a series of paywalled gardens, limiting access to knowledge and creativity.
When an image is created using AI, there are often signs that give it away. Sometimes, the creator will mention it in the image’s description, while other times, there may be abnormalities in the image itself. These abnormalities could include missing features, strange green flowers, or oddly smooth backgrounds. However, AI detection tools are now available to determine the extent to which AI has been used in creating an image.
One such tool is Maybe’s AI Art Detector. Google’s AI arm, DeepMind, has also introduced a new SynthID tool. This tool has two capabilities: watermarking AI-generated images and recognizing them. With the help of these detection tools, it is possible to differentiate between images artificially created with AI and those indistinguishable from reality.
Despite US sanctions, China is seeking to develop its own high-bandwidth memory (HBM) chips for AI processors to achieve semiconductor self-sufficiency. ChangXin Memory Technologies (CXMT), China’s top dynamic random access memory (DRAM) maker, is seen as the country’s best hope for producing HBMs, but it may take up to four years to bring products to market. According to industry sources, Chinese chip makers may have to use less advanced technologies to manufacture DRAM chips. SK Hynix, Samsung Electronics, and Micron Technology are the global HBM production leaders.
Demand for HBM chips is expected to grow by nearly 60% in 2023 as they are the preferred solution for overcoming memory transfer speed restrictions. SK Hynix recently announced the development of HBM3E, the next generation of high-end DRAMs for AI applications, with mass production scheduled for the first half of 2024. Despite the technological challenges, industry insiders believe China can produce its own versions of HBM chips without the latest equipment.
Major media companies, including Disney, The New York Times, and CNN, are taking measures to block OpenAI’s ChatGPT from accessing their content. These companies are adding code to their websites to prevent ChatGPT’s web crawler, GPTbot, from scanning their content. Other outlets and publishers such as Bloomberg, The Washington Post, and Vox Media also take similar protective steps. The media companies are concerned about the unauthorized scraping of their copyrighted content by GPTbot.
This scraping is used to train ChatGPT’s generative chatbot to respond, but it can cause harm and infringe on copyright. Multiple media companies are considering lawsuits against OpenAI for copyright infringement, and comedian Sarah Silverman and two novelists have already sued the company for scraping their works without consent or compensation. OpenAI is also facing scrutiny from the Federal Trade Commission over its data collection system. In addition, the European Union is developing regulations for artificial intelligence. Protecting content from scraping has become a significant concern for media companies, as they face ongoing labor disputes related to artificial intelligence and possible job replacement.
Snapchat has released a new feature called Dreams, which uses generative AI to let users take themed selfies. The app’s Memories section feature allows users to create “fantastical” images and try out new identities, such as a mermaid or a renaissance-era royal. Dreams offers eight themed images at launch, but additional options can be purchased for $1 each. Users can upload selfies capturing various poses and select the theme they want to create AI-generated selfies. The feature will soon allow users to add friends to the image if they have opted in.
Dreams was initially launched in Australia and New Zealand and will be rolled out to other countries in the coming weeks. This feature follows Snapchat’s “My AI” chatbot, which launched earlier this year and is powered by the AI language model ChatGPT.