A few minutes every morning is all you need.
Stay up to date on the world's Headlines and Human Stories. It's fun, it's factual, it's fluff-free.
The AI revolution is fully upon us. At this point, those AI image generators are old news; people still use them, but they’ve lost the trendiness they once had. Now, the internet is getting on board with an entirely new AI phenomenon: chatbots.
Yes, chatbots have been around for a while. That’s probably who you message when you need product support online. But, you might be surprised by how far back the history of chatbots actually goes, with a computer program called “Eliza” being invented back in the 1960s as a kind of computerized therapist. Eliza was pretty basic and didn’t really create the best therapy sessions. But she’s considered to be the first AI chatbot.
But the bright and shiny new AI bots are a little different. We’re talking about, in particular, ChatGPT.
Last week, OpenAI released ChatGPT for public use. This is the same organization responsible for the DALL-E AI image generator. In a tweet last Thursday, OpenAI’s CEO Sam Altman said, “There is a lot more demand for ChatGPT than we expected; we are working to add more capacity.”
It’s kind of scary how almost perfect ChatGPT seems. This thing can write full-on essays that are impressive to read. Since it was released, people have been feeding the system exam questions, with the bot responding in essays that academics say would be more than acceptable if handed in by an undergrad. Computer programmers, too, are testing ChatGPT with coding challenges in lesser-known programming languages. It can solve them in seconds.
At the same time, though, sometimes ChatGPT can be straight-up wrong, even if it sounds right. Its developers have admitted that it “sometimes writes plausible-sounding but incorrect or nonsensical answers.” For example, ChatGPT gave one Twitter user a very detailed but very incorrect response to an algebra equation that the user fed it.
When it was released, OpenAI said, “The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.” We’ll see.