This week in data: AI stack tricks, generative AI adoption, the future of composability (and more)
This week, Bruno talks everything from data management and composability, to generative AI adoption trends, to standing out in sales.Read More
As drone traffic increases, researchers turn to AI to help avoid collisions
Autonomous drone aircraft traffic in uncontrolled airspace below 400 feet altitude is expected to substantially increase in the next few years. Experts anticipate a fleet of nearly 1 million commercial uncrewed aircraft systems (UAS) in the U.S. by 2027, engaging in tasks like package delivery, traffic monitoring, and emergency assistance.
Israel’s Failure to Stop the Hamas Attack Shows the Danger of Too Much Surveillance
Hundreds dead, thousands wounded—Hamas’ surprise attack on Israel shows the limits of even the most advanced and invasive surveillance dragnets as full-scale war erupts.
Study: Digital watermark protections can be easily bypassed
Perhaps the most chilling aspect of AI is its capacity to generate deepfake images.
Ten years in: Deep learning changed computer vision, but the classical elements still stand
Deep learning hasn’t (yet) rendered classical computer vision obsolete. Why some challenges are still best solved with traditional algorithms.Read More
12 Best Umbrellas (2023): Windproof, Cheap, Tiny, and Clear Bubble
These picks will protect you from showers, withstand the wind, and hold up for the long haul.
Researchers turn to Harry Potter to make AI forget about copyrighted material
Their unlearning technique may also be more effective for fictional texts than non-fiction, since fictional worlds contain more unique…Read More
23andMe User Data Stolen in Targeted Attack on Ashkenazi Jews
At least a million data points from 23andMe accounts appear to have been exposed on BreachForums. While the scale of the campaign is unknown, 23andMe says it’s working to verify the data.
Two experiments make a case for using deepfakes in training videos
Watching a training video featuring a deepfake version of yourself, as opposed to a clip featuring somebody else, makes learning faster, easier and more fun, according to new research led by the REVEAL research center at the University of Bath.
StreamingLLM shows how one token can keep AI models running smoothly indefinitely
An innovative solution for maintaining LLM performance once the amount of information in a conversation ballooned past the number of tokens…Read More