Fact Check Enhancements

Here is a short list of recent notable AI developments that we think professional writers and publishers should be paying attention to:

AI Share of Web Articles Grows. A study by the SEO firm Graphite concluded that new content on the web is created roughly equally  by humans and by AI. TL/DR: Most of this content is spammy and marketing oriented, not the type of high stakes writing that Symbolic enables. With that said, its clear that AI is making inroads in professional content of all types. 

Apple Sued for Unauthorized Training. An industry group has sued Apple, claiming that it used scientific and medical books to train models without authorization, following its successful lawsuit against Anthropic for using other books in its training models 

OpenAI Embeds Apps in ChatGPT. OpenAi is allowing third party apps to integrate with ChatGPT, facilitating a straight through process between a  chat session and the app. Will chat become the primary UI of the AI era, like search did of the web? (we don't think so).

And here's what's new in Symbolic: 

Fact Check: Multiple Claims in a Statement

Symbolics' Fact Check is the most comprehensive solution available, and we continue to add enhancements to give users more accuracy, control, and precision. 

Fact Check now takes the next step by independently evaluating each claim in a statement. Symbolic isolates statements from the draft and identifies all possible claims made within that statement. Symbolic then independently checks each claim. A "claim" becomes a "fact" when it is verified and supported. 

The new Fact Check user interaction shows the atomic units clearly; statements are set off as "parent" entities, with their subsidiary "child" claims listed below. The purpose of this approach is to allow our users to evaluate information in their drafts at a more granular level. 

As an example, a statement that "Apple and Amazon report earnings today" could be partially true. Symbolic will now test each part independently, allowing our users to draw their own conclusions. As always, support is gathered hierarchically from a) research gathered by the user, b) pre-designated Trusted Sources, and 3) the web.

Clicking into a statement will bring up a new modal which will show the statement, its associated claims, and an explanation for why each claim is supported, unsupported or refuted. Clicking into a claim shows the user each source that formed the basis of Symbolics judgement. Symbolic also highlight excerpts from each source, making it easy to drill deeper into the reasoning.

Factcheck: Quick View

Fact Check now provides a quick view hover over at the claims level. In the document, hovering over any statement shows the claims associated with it, color coded to show their status as supported, unsupported or refuted.

That's it from us this week. We are always working to improve Symbolic and make your experience more efficient and enjoyable. We would love to hear your feedback and any suggestions you may have for future updates. Don't hesitate to reach out to team@symbolic.ai.