Key to R1’s success was smoothing, a technique that makes AI models more efficient. It works by getting a larger model to tutor a smaller model: you run the teacher model over many instances and record the responses, and reward the student model because it copies those responses as closely as possible, so that it gets a compressed version of the teacher’s knowledge. — Kiwi Chan
10. Sycophancy
As people around the world spend in ever-increasing amounts Communicating with Chatbots Like ChatGPT, chatbot makers are struggling to adopt the tone and “personality” that models should adopt. In April, Openei admitted that it struck the wrong balance between helpful and derisive, saying that a new update had also labeled GPT-4O as sycophantic. It’s not just annoying to suck you in – it can mislead users by reinforcing their false beliefs and spreading misinformation. So consider your reminder to take everything — yes, everything — with a pinch of salt. rhrhiannon Williams
11. Slope
If there’s one AI-related term that’s completely escaped idiots and entered the public consciousness, it’s “slope.” The word itself is old (think pig feed), but now “slope” is commonly used to refer to low-effort, mass-produced content produced by AI, often optimized for online traffic. Many people even use it as a shorthand for any AI-infused content. It’s felt inevitable in the last year: we’ve marinated in it, since Fake biography to Crab Jesus Pictures Unreal human-animal hybrids Videos
But people are having fun with it too. The term’s derisive flexibility has made it easy for Internet users to slap on all sorts of words with a suffix to describe anything that’s ridiculously trivial: think “work slop” or “friend slop.” As the hype cycle resets, “Slope” marks a cultural calculus about what we trust, what we value as creative labor, and what it means to be surrounded by something that was made for engagement rather than expression. — Kiwi Chan
12. Physical intelligence
Have you had access to hypnosis? Video A humanoid robot pouring dishes in a dark, gray-scale kitchen from earlier this year? This gives rise to the idea of ​​physical intelligence: the idea that advances in AI can help robots better navigate the physical world.
It’s true that robots have been able to learn new tasks faster than ever before, everywhere from operating rooms to warehouses. Self-driving car companies have also seen improvements in how they simulate roads. That said, it’s still wise to be skeptical that AI has revolutionized the field. For example, consider that many robots advertised as butlers in your home are doing the majority of their work thanks to remote operators in the Philippines.
The road ahead for physical intelligence is also strange. Large language models train on text, which is abundant on the Internet, but robots learn more from videos of people working. That’s why data from the robot company suggested in September that it would Pay people Working yourself to work in your apartments. Will you sign up? James O’Donnell
13. Fair Use
The AI ​​model is trained by consuming millions of words and images on the Internet, including copyrighted works by artists and authors. AI companies argue that this is “fair use”. This is a legal theory that allows you to use copyrighted material without permission if you transform it into something new that doesn’t compete with the original. The courts are starting to weigh in. In June, the book library was ruled fair use of its AI model Claude’s Anthropic training because the technology was “.Extreme change.
In the same month, Meta scored Similar winsbut only because writers couldn’t show that the company’s literary buffet cut into their paychecks. As copyright battles rage, some creators are cashing in on the invitation. In December, Disney signed a Splashy deal With Openai, the AI ​​video platform lets users of Sura create videos featuring over 200 characters from Disney franchises. Meanwhile, there are governments around the world Rewriting the Copyright Rules Material for gosling machines. Is there AI training on fair use of copyrighted works? As with any billion-dollar legal question, It depends. memiclelle less
14. Live
Just a few years ago, an entire industry was built around helping websites rank higher in search results (well, only in Google). Now search engine optimization (SEO) is giving way to geo-generative engine optimization as the AI ​​boom forces brands and businesses to turn to AI to maximize their visibility, whether it’s in AI-enhanced search results like Google’s AI Overview or in response to LLMs. No wonder they’re out. We already know that news companies have experimented Great drop In search-driven web traffic, and AI companies are working on ways to cut out the middleman and allow their users to visit sites directly within their platforms. It’s time to adapt or die. rhrhiannon Williams