-
Sentiment analysis shifting perspectives over this week of news
Inspiration
What matters now? How we show up in public, with each other.
This week, to synthesize the REAL deluge of news I used a mix of research & analysis tools to build https://realnews.lovable.app This allows me to see the news more clearly and plainly from a wider POV, reading a massive amount without getting lost in any one particular lens or mindset, and also lets me dissect a bit of what's happening vs. what is being talked about in the top media sites (US & Global). My goal is to remove some bias and reflect on stories with a human, comparative approach before making some of this work available for interactive, dynamic forms in newsrooms, for new types of newsreels & data-driven storytelling. It also lets me see around blindspots to hear real, human sentiments from around the world.
What it does
Gemini, when asked to synthesize this week's news, reflected a great deal of concern that the news it was reading could not be REAL, and struggled to make sense of the deluge of files & news that seemed to go against things that were programmed in its systems to be not SFW or otherwise inappropriate, as they seemed too grotesque to be true. Yet, we know this deluge of evidence has been a long time coming, decades of unraveling to reach what Gemini called the Breaking Dam, the Cumulative Collapse (the AI's words, not mine).
Gemini, when assessing the social media sentiment of the US across many social networks, outlined what it saw as 3 pillars or columns of a dam all breaking at the same time due to the moral, social and survival outrage felt as everyday humans reach their breaking point with unaffordable food, medicine and housing while national treasures are torn down and sold off for pennies to enrich people who are already billionaires, leaving normal people hungry, angry, over it. So I started making OVER IT hats. KEEP IT REAL hats. YOU ARE REAL hats. REDACTED leggings and mugs. Ways to move past this, paired with the news stories compared using AI tools to do what they do best: synthesize vast amounts of data to find patterns.
How I built it
As a part of this work, I've ended up reading thousands of social media posts across platforms I rarely visit - and everyone is OVER IT; over giving attention to terrible people & atrocities. I'm slowly turning those sentiments into new hats & wearables, gifts to reset our common narratives and set ourselves on a new path. This work pulls from a wide mix of news sources combined with deep research & analysis tools; I'll share more on this methodology in a future video. I built out a Keep it Real brand as well to express how some of us are Over It. This one chart tells that story effectively as support erodes across all groups, including a formerly core base. REAL is set up as a video channel with stories to come that synthesize news sources; more on that coming soon. I do this work with a handful of mediamakers and media companies as a part of my consulting work.
Challenges I ran into
Fact checking sentiment analysis and sites that I rarely or never visit to understand extreme POVs.
Accomplishments that I'm proud of
Built a store, designed new products along with launched the news analysis site Real News in 24 hours.
What I learned
Promptip: if you're trying these kinds of experiments on your own, always ask for citable sources. Many tools will try to synthesize data that you do not have access to (Gemini/Google mentioned "patriot forums" for example along with websites I do not use) so I spent quite a few hours going through those citations to understand where this data was coming from and how reliable it is vs. what the AI shows that reality to be. This took significant prompting time to require the LLM/AI tools to verify all citations with links and then triple check data veracity. Most AI tools do a terrible job with news or current events analysis so always use a deep research method combined with citable sources then check your own data. I may have missed something in these resources; this is a V1 and I'll be continually checking and rechecking every graphic, data set, report and analysis as I do not trust any particular tool to get it 100% right on the first attempt. Expect to go through analysis with a fine tooth comb as a critical-thinking human if you are trying these experiments at home. Never offload your critical thinking skills.
Log in or sign up for Devpost to join the conversation.