Connect with us


Summari is building AI tools that people can trust 

Summari’s focus on trust and accuracy ensures they are prepared to handle cases where AI gets it wrong to deliver a reliable product for publishers and readers.

summari link previews on a purple background
Image: KnowTechie

ChatGPT is the newest star of the tech world, but if you play around with the bot, you might notice that it isn’t always right. Sometimes, it is confidently very, very wrong.

In the disclaimers, before users can start interacting with the bot, maker OpenAI warns the user that ChatGPT “sometimes writes plausible-sounding but incorrect or nonsensical answers.”

That is a little disconcerting. But why does it generate incorrect answers?

Just how good is generative AI today?

Well, generative AI can be very creative, sometimes too clever, and the guardrails put up by the makers of AI tools like ChatGPT often aren’t enough to ensure that the model doesn’t publish misinformation. 

In the generative AI space, Summari is tackling generative AI’s notoriously unwieldy creativity with full force. You might have noticed that links in KnowTechie articles look slightly different these days.

knowtechie web page with summari links displayed

That’s because KnowTechie implemented a product from Summari called Link Previews. So when you hover over a link on a desktop or tap the icon next to it on mobile, the summary that pops up is written by AI.

Don’t panic; Summari has an extensive system of checks and guardrails to ensure that all the content in the Previews window is accurate and true. 

How does Summari keep AI-generated writing in check?

Summari is well aware that AI models frequently generate text that contains errors, and they know that for publishers, reader trust is paramount.

So, Summari focuses on programmatically identifying and fixing the errors generated by the AI to deliver the most accurate summaries.

Its technology, built on thousands of hours of human summarization work, reads text and identifies core themes that will be included for summarization.

Summari example of AI in writing
Image: KnowTechie

The summary is generated based on these themes, is checked for errors (or hallucinations in AI vocabulary), and any identified errors are corrected.

This is all done programmatically and makes Summari summaries so reliable and helpful.

It’s what’s happening behind the scenes that sets Summari apart in generative AI.

Link Previews don’t just deliver trustworthy information; they improve the reader experience, which translates into improved SEO metrics.


Page views


Time on site

A/B testing and Google Analytics confirm Link Previews significantly impact dwell time, bounce rate, and page views. More time on the page means more ad impressions and higher revenue. 

When you hover over a link in a KnowTechie article, you can rest assured that the information is correct. That’s because it’s been rigorously evaluated.

Summari works with large and small publishers to upgrade links into engaging, informative, and accurate previews.

Therefore, it does it without the fear that it might publish misinformation while delivering immediate SEO results.

Final thoughts

summari link previews

The revenue upside from Link Previews with no impact on site performance is a huge win. Generative AI tools are excellent, but they aren’t very useful if the output can’t be trusted.

Summari’s focus on trust and accuracy ensures they are prepared to handle cases where AI gets it wrong to deliver a reliable product for publishers and readers.

Editor’s Note: This article was written by Isabelle Lee, who serves as the Head of Content at Summari.

Have any thoughts on this? Drop us a line below in the comments, or carry the discussion over to our Twitter or Facebook.

Editors’ Recommendations:

Follow us on Flipboard, Google News, or Apple News

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

More in Internet