Is ChatGPT a Disruptor or a Distraction?
OpenAI's latest breakthrough, ChatGPT, is raising eyebrows and hopes.
Welcome, everyone, to the latest debate. Two weeks ago, a new artificial intelligence program, ChatGPT, made its debut online. This project from the OpenAI research lab can write essays and carry on convincing written conversation. Does ChatGPT represent a breakthrough that will spawn new businesses? Or is it more of a gimmick? Jon Fortt is here argue both sides in the latest On the Other Hand.
“ChatGPT is a disruptor and a game changer for business.
Communication — the ability to marshal facts, organize ideas, stir passions — is now something software can do. Computers have achieved a sort of creativity. How did we get here? Seven years ago, Sam Altman, Elon Musk and others pooled more than a billion dollars to research AI to benefit humanity. Three years ago, Microsoft threw in another billion. This is important because it comes in the same year that OpenAI released DALL-E 2, which generates digital images that seem to be photographs or drawings or paintings created by a human. So what's the business application? My bet is on sales, to start. Soon a company's most effective emailed sales overtures will be training material for a ChatGPT-like bot that scours a potential customer's website and crafts a custom pitch. Humans will just swoop in to close the deal. Another use: An AI bot should be able to look at sales and inventory data and quickly craft a narrative that highlights trends that a human wouldn't quickly find. Conversational AI is a tool to help us learn faster. Apply it the right way, and there are billions to be made.”
What about ChatGPT's accuracy problems? Does it know how to say it doesn't know?
“On the other hand, ChatGPT is already getting overhyped.
How do I know? I tried it! I asked it to write this On the Other Hand segment — and I'm happy to report it did a horrible job. Then I asked it to write a Shakespearean sonnet and it performed like a gifted eighth grader who doesn't quite understand iambic pentameter yet. Here's the thing: The problem isn't that ChatGPT is imperfect. Aren't we all? The problem is it betrays zero self doubt. And it doesn't show its work. As a source of information it's a bit of a black box, unlike Wikipedia, which shows you references; or Google, which shows you links. At least with those it's easier to assess the quality of information you're getting. ChatGPT has value in that it commoditizes bland communication, in the same way early business websites eliminated low-level stock brokers and travel agents. But for it to reach that level of business impact where it can disrupt today's information giants, it needs to be more transparent. Ironically, the parlor trick aspect of ChatGPT requires that it pretend to be an expert. But its value will be limited until it gets better at asking for help with what it doesn't know.”
What do you think? Which side do you find more convincing, and why? Watch how it played out on Squawk Box below: