Last year I wrote about how search was changing with AI. Now SearchGPT by OpenAI has been released as a prototype. I was granted early access so wanted to share what it looks like and what this may mean for news organisations. 

TLDR: It’s in the test phase where three different searches brought up three different iterations of what an answer could look like. 

Below you see three screens. The search function starts very much in the same way as ChatGPT with an open box at  the bottom of the page. As you hit “return,” an answer is produced. My first search was a few days before the election, and I was curious to see how it handled a question around polling.

The answer is cited at various points (see first screenshot) which have a link including a brief description on hover over (see second screenshot). There is also a list of sources illustrated by a few icons at the bottom of the text. When you click on these, you get to the screenshot on the right, a clear list of sources in a pop up.

Each link takes you out to a new window:

I then thought I’d try something a little more contentious news-wise and asked who would win the election. I was surprised to see a different format appeared. In this format, citations are given with no hover over, and there is simply a button at the end called “sources” but no logos.

I then thought I would try a completely different search around a hard-hitting news story — the floods in Spain. This answer is much more structured, with subheads and bullets. Again there are mid-text citations with hover overs and logos with the sources.

Clearly this is in trial mode so we’re unlikely to know what the final will look like. And with any consumer tech product, it will be iterated upon constantly.

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .