IBM Patents Technology to Catch Errors in Self-Teaching Software as It Learns
AI models are only as trustworthy as the data they learn from — and right now, there's no standard way to prove…
The unglamorous machinery underneath: training methods, inference speedups, model compression, and the scheduling tricks that make AI affordable to run.
AI models are only as trustworthy as the data they learn from — and right now, there's no standard way to prove…
Training an AI model usually means hauling enormous amounts of raw data to a central server. Red Hat's new patent flips that…
Every photo app has a 'delete duplicates' button, but Qualcomm is patenting something different: an AI that looks at a whole batch…
AI models spend a huge chunk of their processing time on a single operation called 'attention' — and Intel has patented a…
Salesforce has patented a way to teach AI about human relationships and personal traits — by inventing an entirely fictional social network…
AMD has filed a patent for a system that figures out the optimal number of "guesses" a small AI model should make…
Running a heavy AI model on a phone is slow and drains the battery fast. Sony's new patent describes a way to…
IBM is patenting a system that watches you practice a skill, grades your performance in real time, then finds someone who does…
When you train an AI model, not all examples are equally useful — but most training methods treat them as if they…
Training a giant AI model normally requires thousands of GPUs crammed into a single building. Microsoft's new patent describes a way to…
What if an AI could tell when it's confused by incoming data — and automatically simplify its own internal math to compensate?…
Most AI models read information in one direction — layer by layer, like a factory assembly line. Google's new patent describes a…
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.