In-depth articles with more detailed investigations, interesting thought pieces, and more.
A new investigation by Alex Reisner for The Atlantic reveals that major tech companies have used dialogue from over 53,000 movies and 85,000 TV episodes to train their AI systems. The data, sourced from OpenSubtitles.org, includes content from acclaimed series like “The Simpsons,” “The Wire,” and “Breaking Bad,” as well as Oscar-nominated films from 1950 …
Jakob Uszkoreit, one of the authors of the influential 2017 paper “Attention Is All You Need,” reveals insights about the development of transformer technology in an interview with Ars Technica’s Benj Edwards. The paper, written by eight Google researchers, introduced the transformer architecture that now powers most modern AI systems including ChatGPT and various audio …
The debate over artificial intelligence in creative writing has intensified after NaNoWriMo announced it would allow AI usage in its annual writing challenge. Writer Savannah Cordova, in an article for VentureBeat, argues that while AI can serve as a limited auxiliary tool for tasks like brainstorming or spelling, it should not be used for core …
Spotify is facing a growing problem of AI-generated fake music being uploaded to legitimate artists’ pages, Elizabeth Lopatto reports for The Verge. The fraudulent uploads target musicians with single-word names like Health, Annie, and Standards, appearing on their verified pages through music distributors. Industry expert Andrew Batey estimates that this type of streaming fraud costs …
Newly revealed emails from a lawsuit filed by Elon Musk against OpenAI expose the internal conflicts that shaped the artificial intelligence company’s early years. The correspondence, dating from 2015 to 2019, details discussions between co-founders Elon Musk, Sam Altman, and other key figures about the organization’s structure and leadership. The documents show that OpenAI initially …
Currently, there are two main trends in the AI market: AI as a product and AI as a feature. AI as a product mainly comes from startups like OpenAI, Anthropic, Ideogram, or Runway. You create an account and usually pay a monthly subscription fee to use these tools. Often, there is a free trial access …
Leading AI companies are creating new ways to evaluate increasingly sophisticated AI models as current testing methods prove inadequate. According to Cristina Criddle’s report in the Financial Times, companies like OpenAI, Microsoft, Meta, and Anthropic are developing internal benchmarks because their latest AI systems achieve over 90% accuracy on existing public tests. Meta’s generative AI …
A comprehensive new approach to measuring returns on generative AI investments has emerged, addressing widespread challenges in quantifying AI’s business impact. According to a KPMG survey cited by VentureBeat’s James Thomason, while 78% of C-suite leaders express confidence in generative AI’s ROI, most companies struggle to measure its actual value. The article presents a detailed …
Dr. Ashley Beecy, Medical Director of AI Operations at New York-Presbyterian Hospital, outlines a measured approach to implementing artificial intelligence in healthcare settings. In a VentureBeat article by Michael Trestman, Dr. Beecy emphasizes three key elements for successful AI adoption: establishing good governance frameworks, following a needs-driven approach, and maintaining transparency to build trust. She …
A recent study by Lakera’s team demonstrates how GPT-4V can be manipulated through visual prompt injection attacks. As detailed by author Daniel Timbrell in his article, these attacks involve embedding text instructions within images to make AI models ignore their original programming or perform unintended actions. During Lakera’s internal hackathon, researchers successfully tested several methods, …