The OpenAI company launched on 1eh February, a paid version of its chatbot ChatGPT. Called ChatGPT Plus, the service is currently reserved for the US and available on a waiting list. For 20 monthly dollars (about 18 euros), it gives access to an accelerated version of the robot that answers questions faster. It is also available at any time, when its free counterpart is often saturated with millions of connections, so it sometimes gets blocked.
Two months after its release, ChatGPT is attracting interest from the world of media and technology: Buzzfeed intends to work with OpenAI, and above all, Microsoft, one of its historic shareholders, has provided new funding of $10 billion. .
Windows developer has integrated since 2021 two OpenAI tools in its software catalog: a computer code writing tool based on GPT-3, and within its Microsoft Designer software (currently available on the waiting list), a code generator image based on Dall-E. According to The Information website and Bloomberg news agency, Microsoft is now working to integrate ChatGPT into its Bing search engine to improve its natural language responses. However, as the head of OpenAI Sam Altman himself points out, this chatbot is far from reliable: in December he explained that “it would be a mistake to use it for something important”.
“Red Alert” at Google
According to a study by the bank UBS, ChatGPT would have been used by one hundred million people during the month of January, which would make the service “to the fastest growth in the history of the Internet”. Sundar Pitchai, Google boss, himself admitted this during a meeting with his employees two weeks after the launch of ChatGPT: 2023 could be a “point of inflection” for the use of textual artificial intelligence (AI) intended for Internet research.
Questioned on this occasion by employees who wondered about “missed opportunity” to launch a competitor to ChatGPT in Google’s colors, Sundar Pitchai defended himself by recalling that a chatbot represented a significant reputational risk for a large company, and emphasized that Google should be maneuvered more carefully than a start-up like OpenAI.
However, in an investigation published on January 31, the CNBC TV channel reveals that the American company is currently testing a conversational robot similar to ChatGPT, called “apprentice bard”, or “apprentice bard”.
According to CNBC, Google is testing this tool’s performance against ChatGPT and is studying, on the home page of its search engine, the best place to include an area where you can write questions to the bot. According to New York Timesthe emergence of ChatGPT prompted Google’s management to issue a “red alert” symbolizing imminent danger to their business model.
Register AI-generated tasks
While waiting for a possible chatbot with the Google logo, OpenAI continues to work on making its own more consensual. This includes the release of a tool to detect texts created by AI, to extinguish the biggest fire that ChatGPT has lit so far: the appearance of tasks generated by AI in classes and universities, causing an outcry in the educational body, even in the lecture halls in Sciences Po, which prohibits its use under penalty of exclusion.
OpenAI’s detection tool, after being fed text, rates the likelihood of it being AI-generated on a scale of 1 to 5. However, the company warns that its detection capabilities are not perfect: some texts are not identified and others are erroneous reported. Additionally, this tool, currently designed to identify texts written in English, works best on those with more than 1,000 words.