The excitement around ChatGPT, an easy-to-use AI chatbot that can deliver an essay or computer code on demand and in seconds, has sent schools into a panic and Big Tech green with envy.
ChatGPT’s potential impact on society remains complicated and unclear, even as its creator announced a paid subscription version in the United States on Wednesday.
Here’s a closer look at what ChatGPT is (and isn’t):
Is this a turning point?
The November launch of ChatGPT by the Californian company OpenAI may well be remembered as a turning point in introducing a new wave of artificial intelligence to the general public.
What’s less clear is whether ChatGPT is really a breakthrough, with some critics calling it a brilliant PR move that helped OpenAI garner billions of dollars in investment from Microsoft.
Yann LeCun, chief scientist for artificial intelligence at Meta and a professor at New York University, believes that “ChatGPT is not a particularly interesting scientific breakthrough,” calling the app an “eye-catching demo” built by talented engineers.
LeCun, speaking to the Big Technology Podcast, said that ChatGPT doesn’t have “any internal models of the world” and is simply churning out “one word after another” based on inputs and patterns found on the internet.
“When working with these AI models, you have to remember that they are slot machines, not calculators,” said Haomiao Huang of Kleiner Perkins, the Silicon Valley venture capital firm.
“Every time you ask a question and pull your arm, you get an answer that could be wonderful…or not…Failures can be extremely unpredictable,” Huang wrote on Ars Technica, the technology news website.
just like google
ChatGPT is powered by an AI language model that is almost three years old, OpenAI’s GPT-3, and the chatbot uses only a fraction of its capacity.
The real revolution is human chat, said Jason Davis, a research professor at Syracuse University.
“It’s familiar, it’s conversational, and guess what? It’s like putting a search request on Google,” she said.
ChatGPT’s rock-star-like success even surprised its creators at OpenAI, which received billions in new funding from Microsoft in January.
“Given the magnitude of the economic impact we expect here, more gradual is better,” OpenAI CEO Sam Altman said in an interview with StrictlyVC, a newsletter.
“We released GPT-3 almost three years ago… so the incremental upgrade from that to ChatGPT, I felt like it should have been predictable and I want to do more introspection as to why I was a little bit miscalibrated on that,” he said.
The risk, Altman added, was surprising the public and lawmakers and on Tuesday his company unveiled a tool to detect AI-generated text amid concerns from teachers that students may rely on artificial intelligence to do their homework.
Now what?
From lawyers to speechwriters, coders to journalists, everyone is waiting breathlessly to feel the disruption caused by ChatGPT. OpenAI has just released a paid version of the chatbot – $20 per month for faster and improved service.
For now, officially, the first significant application of the OpenAI technology will be for Microsoft software products.
Although details are scant, most assume that ChatGPT-like capabilities will appear in the Bing search engine and Office suite.
“Think Microsoft Word. I don’t have to write an essay or an article, I just have to tell Microsoft Word what I want to write with a prompt,” Davis said.
He believes that TikTok and Twitter influencers will be the first to adopt this so-called generative AI, as going viral requires a lot of content and ChatGPT can take care of that in no time.
Of course, this raises the specter of disinformation and spamming on an industrial scale.
For now, Davis said, ChatGPT’s reach is severely limited by computing power, but once it scales up, the potential opportunities and dangers will grow exponentially.
And just like the ever-imminent arrival of driverless cars never quite happens, experts disagree on whether it’s a matter of months or years.
Ridiculous
LeCun said that Meta and Google have refrained from releasing an AI as powerful as ChatGPT for fear of ridicule and backlash.
Quieter releases of language-based bots, like Meta’s Blenderbot or Microsoft’s Tay, for example, quickly proved capable of generating racist or inappropriate content.
Tech giants have to think long and hard before launching something “that is going to be bullshit” and disappoint, he said.
© Agence France-Presse
