Yann LeCun, the chief scientist at Meta (formerly Facebook) took to social media today to discuss the public perception of ChatGPT, the artificial intelligence (AI) platform built by Microsoft-backed AI research firm OpenAI. Speaking to reporters and executives at the online discussion series the Collective[i] Forecast, organized by Collective[i], Lecun said that the chatbot, which has enthralled many, is “not particularly innovative”.
“In terms of underlying techniques, ChatGPT is not particularly innovative,” Lecun said, according to a report by ZDNet. He later tweeted, while citing the ZDNet article, that ChatGPT and other large languages are the results of decades of contributions from various people and no one AI lab is significantly ahead of others.
The title is blunt, but the article conveys what I said on the @collectivei Q&A about progress in AI.— Yann LeCun (@ylecun) January 23, 2023
ChatGPT & other LLMs didn't come out of a vacuum & are the results of decades of contributions from various people.
No AI lab is significantly ahead of the others. https://t.co/WtLMrQVnVm
A little later, once his comments had started doing the rounds among the tech community, LeCun clarified that he wasn’t criticizing OpenAI’s work. Instead, he said that the comments were directed toward the public perception of ChatGPT as an “incredibly new, innovative, and unique technological breakthrough that is far ahead of everyone else.”
LeCun is among the first notable voices in the field of AI research to have downplayed how advanced ChatGPT is. OpenAI, which received its third investment from Microsoft this week, has gained attention as people were amazed by ChatGPT’s capability of talking like humans and generating coherent answers to questions.
In fact, Bloomberg reported that ChatGPT’s prowess with natural language generation (NLG) even has AI behemoth Google worried. The company is reportedly looking to fast-track and launch 20 AI products at the Google I/O this year, and even founders Sergey Brin and Larry Page had been brought in to relook at Google’s AI roadmap, according to a report by The New York Times.
To be sure, despite the popularity of its AI tool, Open AI CEO Sam Altman, has warned that ChatGPT can have flaws. On December 11, he tweeted, “ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness.” He added that the tool is in the preview of progress and a lot of work needs to be done on the robustness and truthfulness aspects of the tool. In a recent with StrictlyVC, Altman also addressed the rumours around the highly anticipated GPT-4 model. “People are begging to be disappointed and they will be,” he said and further added, “We don’t have an actual AGI and that’s sort of what’s expected of us.” Altman was answering the question about the size of GPT-4, which many speculate to be up to 100 trillion parameters.