Stanford Institute for Human-Centered Artificial Intelligence (HAI)的动态

This paper took about -- hmmm -- *six* years to get published. Probably our team's record. Getting it out was thanks to the perseverance of Jim Jansen to not give up. The paper itself it based on an algorithm developed by Raghvendra Mall (co-ideated by me and implemented by Mridul Nagpal) that identifies polarized content based on the distribution of semantic indicators -- M- or U-shaped distributions in which low- and high-range observation densities are high. We use the toxicity score (values obtained by Hind Almerekhi) to illustrate this idea, but one could use the same logic with any other semantic or textual indicator, like sentiment, or even other metrics like having simultaneously high like and dislike frequency, to apply this idea. In any case, we apply the algorithm to score 3700 YouTube news videos with 180k comments in total (data collected by Soon-gyo Jung). Empirical findings indicate that polarization is highest with politically sensitive topics (this makes sense and lends credence to our approach) and that high engagement generally co-exists with high polarization, emphasizing the inherent challenge of social media platforms to maximize engagement while minimizing negative interactions. Read full paper in open access: https://lnkd.in/dVCMgTQq #research #openaccess #toxicity #polarization #youtube #detection #algorithms

The news you consume is controlled by your algorithm. ?? Take a moment to let that sink in. Historically, everyone would consume pretty much the same news, through traditional media such as newspapers, magazines, and TV. Sure, there were variations based on personal interests, but broadly speaking, everyone was exposed to the same headlines. Now, here’s a shift to ponder: 71% of 16 to 24-year-olds are using social media to stay abreast of the news. The very nature of these platforms means that the news which resonates most with you is more likely to be presented to you. This insightful piece of research from Ofcom could very well mark a turning point in how we approach public news consumption, especially in the wake of generative AI's rise. The potential impact on our news consumption habits is not just fascinating; it’s a prompt for introspection. The question that could shape the future of news consumption is clear: In an era dominated by personalised content, how will this change affect the way major events and stories are shared and perceived? With an increasing reliance on algorithms, ensuring a diverse and comprehensive view of the world becomes a subtle, yet profound, challenge. The implications for audience engagement and content diversity are significant. Social media is now the main source of news for most people. This reality marks a fundamental shift in how information is disseminated and consumed, reshaping our understanding of the world in ways we are only beginning to grasp. https://lnkd.in/emd_22uZ #NewsConsumption #SocialMedia #SocialImpact

Disinformation on social media is more dangerous than ever, especially with artificial intelligence-generated content shaping political narratives. ?? Lance Hunter explains how AI-driven disinformation can undermine democracy, and therefore is calling for vigilance in evaluating online content. https://bit.ly/4ecR1os

Understanding algorithmic amplification is necessary to explore the processes and tools that empower users so that they can proactively shape and assess the content they engage with. Labels can be a helpful tool in the broader algorithmic transparency toolkit, argue Marlena Wisniak and Luca Belli: https://lnkd.in/eYSRYMen

Hi Everyone, It has been almost five months since I posted on this social media platform. Real life is extremely busy leaving completely no room for social media activities. But as usual, I've not come emptyhanded. I come bearing blogs on my favorite topic of emotions, the parent of emoticons. ?? . [Oh my gosh, the Californian sun is too hot. It has been eleven long years in California, but I need my shades even now] Check out my first three articles on 'Sentiment Analysis- the emotional aspect of social media website, reviews and games' I love your feedback, so don't forget to comment or share. Also, if you're new to me, follow me on https://lnkd.in/gDzSCYwZ as the series has begun recently and in its fledgling stages. https://lnkd.in/gVNAxUck https://lnkd.in/ggSTGVsj https://lnkd.in/gU7Sh2nq

Algorithms on social media platforms can create problems in a variety of ways, as organizations like the Center for Humane Technology have pointed out. But, we can also "teach" algorithms to feed us the types of content that are genuinely useful to us. I recommend taking some time to curate your social media feeds by following the voices you want to hear from and unfollowing those who aren't adding value for you. This isn't to say that you should create an echo chamber. (See Alice Thwaite's excellent work on that topic for more information/analysis.) Find voices that are authentic, honest, and respectful, even if some of their opinions differ from your own. For example, Michael Shellenberger is someone I follow and read to get a take that's often at odds with some of my beliefs but challenges me to think critically about some of the big ethical questions (like free speech versus censorship).

Negative 45% That’s the price of inconsistency that are shown by analytics in my linkdeln for not posting. But that’s just for the people on this professional social media platform. What is it for me? A positive 20%. Why? 1. Took a break to enjoy my hometown and not become mad by the need to post. 2. Learnt that if the previous content is good enough, you can rest back. 3. I don’t need chat gpt for hooks of my content,well atleast not for a fresh piece . Thus the only thing learnt from the countless binge watches in the past week is- Slow is smooth Smooth is fast. #growth #personalbranding

Have you heard about the #deadinternettheory? Because the chances are you're already seeing it in action. At a time when AI-generated content is rampant and bots account for around?half?of all internet traffic, the theory has taken on a whole new meaning in 2024 - leaving many netizens concerned about the future of the web. Check out my article below to find out why bizarre phenomena like Shrimp Jesus are becoming more common than ever, and what this reveals about the state of the internet today. ?? ?? ?? https://lnkd.in/eT59TK_U

This was my Substack comment on the excellent (23 May 2024) historical comparison by Heather Cox Richardson of our current, regressive Supreme Court decisions, to Abraham Lincoln‘s famous “House Divided” speech of June 16, 1858. “These factual misconceptions about the current status of the United States economy are not only troubling, but also existential. When a large majority of citizens of a country do not understand basic facts about their economy, politics and culture because of news outlets, as well as social media influencers, who intentionally obsfucate reality and present an alternative one, the ability of citizens to vote rationally and make reasoned judgements based on objective reality is eliminated.” https://lnkd.in/eZ6YTNKR Sadly, I believe these kinds of misconceptions about politics in our culture, which are shaped more by social media and social media influencers than by Presidential debates or messages political parties / candidates themselves / established mainstream media outlets can communicate, will decisively shape the outcome of our November 2024 presidential election. Spoiler alert: I do not think the outcome will be positive for the causes of human rights, self-determination, or representative democracy. Image attribution: AI visualization via Ideogram: https://lnkd.in/eAtfZpnC Edited with Skitch for iOS (unfortunately no longer available on the US iOS App Store) Learn more about my #ConCW ”Conspiracies and Culture Wars media literacy inquiry project:” https://lnkd.in/e4fqTYV5