Existing AI News Research - Cian - 26/9/23

As a starting point for our project, I decided to research existing examples of AI driven News Aggregators.

Feedly:

Feedly is a news aggregating app which the user can train to filter specific kinds of articles. It functions as an RSS feed reader similar to the new defunct Google Reader. 

The app pulls in news sources from a variety of publications which can be chosen by the user. These can be news websites, scientific journals, and blogs.  

Feedly’s AI Leo can be trained to pick out articles which the user will be most interested in. The AI detects the level of user engagement with different articles by measuring click through rates and read time. 

These features allow for a much more personalised news experience, however it risks putting the user into an echo chamber of hyper specific news sources. 

 

Artefact:

Artefact is a social news app powered by machine learning. It was created by the co-founders of instagram to be an unbiased news source. 

Artefact is powered by a machine learning algorithm which learns from user behaviour, however its not solely user driven. 

The app judges sources based upon journalistic integrity and transparency. The algorithm is allegedly programmed to be editorial and objective, however I could not find hard proof of how this is determined.

 

 

NewsNotFound:

NewsNotFound is an ai powered news website. It advertises itself as fully transparent and unbiased. All articles are AI generated.  

The AI scrapes headlines from public websites and generates articles based on the facts within those articles. It uses GPT-3 and GPT-4 and a specific prompt to generate a full, unbiased news article from the previously generated one.

Personally I have a huge issue with this kind of AI use. The information is being scraped from already existing articles and journalistic reports, therefore the information collected will inherently carry some form of bias. 

While the AI does do a good job of stripping out the facts, I personally feel like it makes the articles lifeless, especially paired with the AI generated header images.


Conclusion:

AI powered news sources are a niche that has been quickly filled by many different kinds of players, however they all leave something to be desired. Whether that's a sense of life, or unbiased journalism, theres room for improvement which we can take advantage of in our project.

Sources:

NewsNotFound.com/white paper/

https://www.news9live.com/technology/app-news/i-used-artifact-the-ai-app-by-instagram-co-creators-so-you-dont-have-to-au137-2059231

https://feedly.com/

Comments

Post a Comment

Popular posts from this blog

Adobe XD Mobile App (Adam)

Image
 I was responsible for the designing of the mobile version for our Citizen Kane news application. It was a little tricky at first as I wasn't too familiar with Adobe XD.     This was the layout of the mobile app .     And this is the XD file with all of the linkages
Read more

In-Class notes, 24th Oct - Research and Prototyping - Cian

  Notes from Phil’s Class on research and prototyping:     RESEARCH:   Research presentation on the 7 th November, worth 30% of final mark!     5 minute presentation, 5 min Q and A.     S lide format is important:   Title   Overview   Aims   Background   Objectives   --All the above within approx 1 min 30 sec   RESEARCH   --Research is the bulk of the presentation.   Next Step   --1 or 2 slides outlining our prototyping plan     Practise, Practise, Practise!! Make sure our timing is consistent.   Avoid using templates for the slides, present information clearly and concisely.     Have backups! Multiple USBs with the PowerPoint would be ideal.     AI should only be PART of the solution.       PROTOTYPING:     Allows us to fail safely, and quickly. Reach a solution faster.     What is it for? - Aim   Who is it for? ...
Read more