What you need to know
- OpenAI has debuted its SearchGPT prototype, which is a temporary limited test that users can sign up for on the waitlist.
- The AI is designed to give "more choice" to users seeking information on the web while being "quick and direct."
- Just recently, Google debuted its new 1.5 Flash AI model for Gemini, which focuses on speed and efficiency to deliver results to users.
OpenAI woke up and chose rivalry as it began a limited test for a new prototype to compete with what Google is known for.
In a blog post, the company explained that SearchGPT has arrived and is a "temporary" prototype search engine that combines the "strength of our AI models." OpenAI aims to solve the issue behind users seeking information on the web, which often requires "multiple attempts" to get the right answer.
SearchGPT is said to advance the company's conversational AI work while also providing "quick and direct" answers to your query. Information is said to be up-to-date and the AI will offer links to fact-check its response. The snapshots of information provided by SearchGPT will include a link to the publication or source in parentheses.
The main source of your results will be featured on the right-most portion of your screen. Beneath, users will find a "Ask follow-up" text bar. Additionally, OpenAI has included a sidebar full of extra results and answers that could offer more insight into your query.
The post adds that users will have the ability to ask a follow-up question. OpenAI states that the AI will use the previously provided information as a building block to continue your hunt.
Those interested can sign up for the waitlist to try OpenAI's SearchGPT prototype. Access to the AI is limited and the company states those selected will receive an email detailing their acceptance.
According to OpenAI, SearchGPT is focused on bringing attention to publishers and creators who can deliver what the user seeks. As such, it has partnered with them to polish the experience and bring "more choice" to searching.
Just recently, Google debuted its new 1.5 Flash model for Gemini, which it proclaims is its "fastest." Optimized for speed and efficiency, Google says its AI features a longer context window (4x over its Pro model) to help Gemini handle complex queries a little easier. Moreover, Google is bringing more fact-checking abilities to Gemini, similar to its approach with AI Overviews.
A small chip — chip is the term Google uses for dynamic information snippets, often looking like a button or small chat bubble, that can pull data from third-party apps and services — will be displayed beside the responses Gemini gives. Interacting with those will display Gemini's sources in case users want to double-check its information.
The model started rolling out on the web and Android devices on July 25.
