It looks like AI fatigue is upon us: OpenAI’s website traffic plummets for the first time since the company launched the ChatGPT viral chatbot in November.
Web analytics company SimilarWeb reported this week, global traffic to OpenAI’s ChatGPT webpage, including desktop and mobile traffic, fell 9.7% in June. In the United States alone, the drop in traffic is estimated at 10.3%. at the same time, unique visitors from around the world to the ChatGPT website fell by 5.7%. SimilarWeb also reports that the time people spent on the webpage also dropped by 8.5%. This would mark the first drop in traffic since OpenAI launched ChatGPT eight months ago.
“Whether OpenAI management is heartbroken over the drop in traffic is debatable. Initially launched as a technology demo, the ChatGPT website primarily serves as a lead-generating lead product for OpenAI, making its technology available for other companies to integrate into their applications,” wrote David Carr, Senior Similar Web Information Officer, Company Report. “Yet the decline in interest not only for ChatGPT but one of its main competitors is a sign that the novelty has worn off for AI chat. Chatbots will need to prove their worth, rather than take it for granted, from now on.
OpenAI did not immediately return Gizmodo’s request for comment on the traffic cut.
The trend is not unprecedented, however, as SimilarWeb said in Can that traffic to the ChatGPT webpage has started to plateau, indicating that the June drop is probably not just a fluke. Bank of America has also found that it’s not just website visits that are declining. CNBC reports that downloads for ChatGPT and Microsoft Bing apps fell 38% from May to June, according to Bank of America analysts citing data from Sensor Tower.
ChatGPT’s erratic website traffic could be a sign that the chatbot’s widespread appeal may be dying out, and while the chatbot has rapidly risen to prominence over the past few months, it has not been without its problems. . A recent study by researchers from the University of California at Irvine and the University of Texas at Arlington found that ChatGPT’s GPT3 training required 185,000 gallons or 700,000 liters of water– a titanic waste that amounts to an average user throwing a large bottle of water on the ground with each conversation. Although it depends on large amounts of water to operate, ChatGPT also depends on human workers, who can be stiff in their paychecks. US-based human contractors, who help label data to help train the chatbot, earn just $15 an hour.