All articles tagged with #algorithms
French President Emmanuel Macron denounced tech giants' claims to defend free speech, insisting platforms must be transparent about how their algorithms work and acknowledging inherent biases; he made the remarks during a New Delhi event on India-France university partnerships, framing the debate within broader EU tech rules and his push to curb youth access to social media.
A Futurism article says AI-generated junk—especially text-to-video—has flooded Facebook and Instagram feeds with disturbing, racist, and bizarre clips, driven by engagement-based algorithms that amplify what users watch; YouTube's Neal Mohan says combatting AI slop is a top priority, while some creators fight back against AI theft, and the situation is expected to worsen as platforms rely on signals to determine what users see.
A Verge opinion piece argues that Instagram chief Adam Mosseri’s focus on authenticity amid AI-powered content misses the core issue: the platform’s algorithm rewards repetitive, easily scalable content—whether AI-made or human-made—making genuine originality hard to sustain without new incentives for real creators.
The article explores how algorithms used in pricing can inadvertently collude and drive up prices, making regulation challenging. It discusses recent research showing that even benign algorithms can lead to high prices and highlights the complexities of regulating algorithmic behavior in markets, emphasizing the subtlety of strategic interactions and the potential for unintended outcomes.
MIT researchers have developed the first provably efficient algorithm for machine learning with symmetric data, which can improve model accuracy and reduce resource requirements by effectively incorporating symmetry into the training process, with applications spanning from drug discovery to climate analysis.
Williams proved a groundbreaking theoretical result showing that algorithms using less space can solve all problems solvable with more time, and established a quantitative gap between the power of space and time, advancing understanding in computational complexity and challenging long-held assumptions.
The article discusses the author's preference for Last.fm over Spotify Wrapped for tracking music listening habits. While Spotify Wrapped is criticized for its AI-driven, marketing-focused presentation, Last.fm offers a straightforward, data-focused archive of listening history. The author appreciates Last.fm's ability to provide meaningful insights into personal music tastes and experiences, contrasting it with Spotify's commercial approach. The piece also highlights the importance of personal agency in music consumption, encouraging listeners to explore beyond algorithm-driven recommendations.
China's Cyberspace Administration is launching a campaign to curb internet echo chambers and discriminatory pricing algorithms. The initiative aims to prevent tech companies from using recommendation algorithms to create 'information cocoons' and to ensure transparency in content ranking. The directive also targets discriminatory pricing in e-commerce, requiring platforms to avoid pricing based on personal factors and to disclose algorithmic processes. Businesses have until the end of the year to comply, with assessments beginning in January.
Researchers have proven that a version of Dijkstra's algorithm, a classic path-finding method developed in 1956, is the best approach for solving the single-source shortest-paths problem across any street grid, assuming worst-case traffic patterns. This algorithm, which is a staple in computer science education, has been shown to be universally optimal, meaning it performs best on every possible network layout. The breakthrough was achieved by focusing on the data structure used in the algorithm, leading to a simpler and more efficient design.
Residents and businesses in Coulsdon, England, have faced Facebook censorship due to the platform's algorithms misinterpreting the 'LSD' in the town's name as a drug reference. Despite complaints, the issue persisted until Facebook's parent company, Meta, acknowledged and fixed the error. This incident highlights ongoing challenges with Facebook's algorithmic content moderation, which has previously led to similar errors, such as censoring posts about Plymouth Hoe and wildfire safety information.
New York is close to passing a law that would restrict social media platforms from using algorithms to promote content to minors, requiring parental consent for algorithmically generated feeds. The legislation also aims to protect children's privacy by limiting data collection and sharing for users under 18. The proposed laws would allow parents to set stricter limits on their children's social media use and notifications. This move has faced significant tech industry lobbying and legal challenges in other states.
A massive leak of 2,500 documents revealing Google's search ranking secrets has sparked outrage in the SEO industry, intensifying the ongoing conflict between SEO experts and Google.
BBC Director General Tim Davie will warn in a speech that British storytelling is at risk of being overshadowed in the global media landscape, with U.S. and Chinese algorithms potentially becoming future taste-makers. He plans to outline strategies for the BBC to transform and renew, including integrating online services, investing in UK content, and boosting commercial income through partnerships. Davie emphasizes the BBC's vital role for the UK and will discuss these plans in detail with the organization's divisions ahead of the Annual Plan.
Democratic lawmakers reintroduced the Justice in Forensic Algorithms Act, which would allow defendants in criminal trials to access the source code of software used to analyze evidence and require the National Institute of Standards and Technology to create testing standards for forensic algorithms. The bill aims to address potential biases and limitations in technology used in criminal proceedings, emphasizing the need for defendants to question the technology and ensuring their due process rights. While the bill does not yet have Republican co-sponsorship, the lawmakers are optimistic about bipartisan support.
Researchers have made a significant breakthrough in the computational efficiency of integer linear programming (ILP) by tightening the upper bound on the covering radius, achieving a dramatic speedup of the overall ILP algorithm. This advancement brings the runtime to (log n)O(n), where n is the number of variables, marking a triumph at the intersection of math, computer science, and geometry. While the new algorithm has not yet been applied to solve logistical problems due to the effort required to update existing programs, it represents a major theoretical advancement with fundamental applications. Further improvements in computational efficiency would necessitate fundamentally new ideas.