In-depth articles with more detailed investigations, interesting thought pieces, and more.
A new AI approach from OpenAI promises significantly better results with longer computation times compared to existing models, reports Michael Nuñez for VentureBeat. OpenAI researcher Noam Brown explained at the TED AI Conference in San Francisco that 20 seconds of AI computation time delivers the same performance boost as scaling up a model 100,000 times. …
A former OpenAI researcher accuses his previous employer of violating copyright laws in developing ChatGPT. Suchir Balaji, who left the company in August 2023, helped collect and organize internet data for the chatbot, as he reveales in interviews with the New York Times. The 25-year-old scientist is convinced that OpenAI’s use of copyrighted data was …
Artificial Intelligence can be better understood through some basic principles of how it thinks and works. Large Language Models (LLM) essentially function like sophisticated text prediction systems that forecast the next word in a sentence based on vast amounts of data, explains Ethan Mollick in his 100th Substack post. These AI systems operate with limited …
Hollywood actor Joseph Gordon-Levitt has issued a stark warning about the dangers of artificial intelligence for the entertainment industry, reports Erich Schwartzel for the Wall Street Journal. Speaking at the Wall Street Journal’s Tech Live conference in Laguna Beach, he criticized AI companies for using movies and TV shows to train their models without fairly …
Media executives are increasingly adopting AI tools for their daily work, though they remain cautious in their approach. According to a Digiday survey at the Digital Publishing Summit in Florida, executives primarily use artificial intelligence for basic tasks like email editing, transcriptions, and research. Some companies have prohibited the use of AI tools until appropriate …
Appen’s 2024 State of AI Report reveals that while generative AI adoption has surged by 17%, companies face significant challenges in data quality and management, VentureBeat writes. The report, based on a survey of over 500 U.S. IT decision-makers, indicates a 10% increase in bottlenecks related to data sourcing, cleaning, and labeling. As AI models …
Anthropic has developed new security assessments for AI models that test their ability to sabotage. In a blog post, the company describes four types of tests: “human decision sabotage,” “code sabotage,” “sandbagging,” and “undermining oversight.” In human decision sabotage, the models try to trick people into making the wrong decisions without arousing suspicion. Code sabotage …
US companies are investing heavily in computers, data centers, and other infrastructure to fuel the AI boom. According to Joseph Politano, investment in fixed assets has increased rapidly to meet the growing demand for computing power. Hundreds of billions of dollars are being poured into high-end computers, data centers, power plants and more. In the …
Schools and universities are increasingly relying on AI detectors to check student work for AI use. According to one survey, about two-thirds of teachers use such tools regularly. As Jackie Davalos and Leon Yin report in Bloomberg Businessweek, at that scale, even low error rates can lead to a large number of false positives. The …
Researchers have developed a new method called “linear-complexity multiplication” (ℒ-Mul) to make calculations in artificial intelligence more efficient. The method replaces complex multiplications with simpler additions, according to Jason Hickey and his team at the Google AI Research Center in Accra. The researchers showed that ℒ-Mul achieves the same accuracy as traditional methods for language …