OpenAI, the brains behind ChatGPT and backed by Microsoft, has just leveled up the chatbot game. They’ve given ChatGPT the superpower to browse the web for the latest scoop.
Until now, ChatGPT was like a time capsule, stuffed with info only up to September 2021. But things are about to get current.
This update means premium users can now quiz ChatGPT on the latest happenings and news. And don’t worry, everyone else will get this feature soon enough.
Earlier this week, OpenAI also teased that ChatGPT will soon be up for a chat – literally, with voice conversations on the horizon.
ChatGPT and its AI cousins have been wowing us with responses that feel eerily human. They’re shaking up how we search online. But until this update, ChatGPT’s know-how was stuck in the past.
For instance, ask the old ChatGPT about recent earthquakes or if Donald Trump’s still around, and you’d get an apologetic “no clue.” This gap has been a buzzkill for some.
Tomas Chamorro-Premuzic, a business psychology prof at University College London, thinks this is a game changer. “Now, ChatGPT is not just another chatbot.
It’s your go-to for the latest news and gossip,” he says. It’s set to siphon off questions normally heading to search engines or news sites.
But Chamorro-Premuzic cautions it’s not all roses. Quick answers are great, but if ChatGPT’s not clear about its sources, misinformation could sneak in. “If it’s just splicing together stuff from the web, how do we know what’s legit?” he points out.
OpenAI’s been in the hot seat with US regulators over ChatGPT spewing fake news. Earlier this year, the Federal Trade Commission wanted the lowdown on how OpenAI keeps reputations safe. OpenAI’s CEO pledged to cooperate.
Why the delay in real-time web searching? Well, there’s the cost – each query reportedly nicks OpenAI a few cents. More importantly, limiting data was a safety net. Without web access, ChatGPT couldn’t parrot harmful content or fall for fake news traps set by shady folks.
ChatGPT itself gave a few reasons for the hold-up: Building language models is a Herculean task, real-time data can be a can of worms, and then there are privacy and ethical hiccups, especially with copyrighted stuff.
The new ChatGPT features are a tightrope walk for AI – to be truly helpful, it needs fewer restrictions. But that opens the door to potential dangers and misuse. It’s a big leap for AI, balancing utility and safety.
