New AI reasoning models, built as neural networks, show striking convergence
When I tried two different reasoning models on the same math puzzle, I was taken aback by how close their answers were. These are just plain-old artificial neural networks - trained on a dataset and a single problem, then left to work it out on their own. Researchers have started to see a pattern: even when the nets differ in size, training data, or the code that runs them, they often land on almost the same conclusion.
Of course, that alignment isn’t a given. Neural nets can wander down very different routes, especially when the task is tough. Still, recent experiments show the outputs beginning to line up, which suggests there might be some regularity in how they handle information.
If the brain-like machinery is converging on the same solutions without extra work, the cost of thinking could be lower than we thought. That observation leads straight into the quote that follows.
"The fact that there's some convergence is really quite striking."
"The fact that there's some convergence is really quite striking." Reasoning models Like many forms of artificial intelligence, the new reasoning models are artificial neural networks: computational tools that learn how to process information when they are given data and a problem to solve. Artificial neural networks have been very successful at many of the tasks that the brain's own neural networks do well -- and in some cases, neuroscientists have discovered that those that perform best do share certain aspects of information processing in the brain. Still, some scientists argued that artificial intelligence was not ready to take on more sophisticated aspects of human intelligence.
"Up until recently, I was among the people saying, 'These models are really good at things like perception and language, but it's still going to be a long ways off until we have neural network models that can do reasoning," Fedorenko says. "Then these large reasoning models emerged and they seem to do much better at a lot of these thinking tasks, like solving math problems and writing pieces of computer code." Andrea Gregor de Varda, a K. Lisa Yang ICoN Center Fellow and a postdoc in Fedorenko's lab, explains that reasoning models work out problems step by step.
"At some point, people realized that models needed to have more space to perform the actual computations that are needed to solve complex problems," he says.
These newer reasoning models aren’t the same as the chat-oriented LLMs we first saw. ChatGPT can spit out an essay in seconds, but it still trips over many math riddles and multi-step logic problems. The fresh generation, by contrast, seems to be taught to pause and walk through a question step by step, much like a person would.
They’re still neural networks at heart, trained on data and on explicit problem statements. Some observers have pointed out a surprising convergence across very different architectures - the researchers themselves called it “quite striking.” How far that convergence goes, and whether it will turn into reliably solid performance in every domain, is still unclear. The longer “thinking” windows hint at handling more intricate tasks, yet early benchmarks give mixed signals.
Critics remind us that speed and fluency don’t automatically mean deeper understanding. So, the field has moved past rapid-fire replies, but we’ll have to see if these reasoning models can consistently beat their predecessors.
Related Reading
LLMs & Generative AI OpenAI says AI saves knowledge workers 40‑80 minutes; use yields five‑fold gains
LLMs & Generative AI Grok Chat: AI for debugging, building, testing web apps with voice and images
LLMs & Generative AI Samsung adds Vision AI Companion, an AI Bixby, to TVs for real‑time queries
Common Questions Answered
What does the article say about convergence among new AI reasoning models?
The article notes that despite differences in size, training data, or underlying code, the new reasoning models—plain‑old artificial neural networks—often arrive at remarkably similar conclusions. Researchers describe this alignment as "striking" and consider it an unexpected pattern across distinct architectures.
How do the newer reasoning models differ from earlier chat‑focused LLMs like ChatGPT?
According to the article, the newer models are trained to pause and work through problems step by step, unlike ChatGPT which can draft essays quickly but still struggles with many math puzzles and multi‑step logic. This shift emphasizes explicit problem statements and a more deliberate reasoning process.
Why is the step‑by‑step approach important for the latest generation of reasoning models?
The step‑by‑step approach helps the models emulate human-like problem solving, allowing them to break down complex tasks into manageable parts. This method improves performance on tasks that require multi‑stage reasoning, such as math puzzles, where earlier models often faltered.
What role do artificial neural networks play in the new reasoning models discussed in the article?
Artificial neural networks serve as the core computational framework for the new reasoning models, learning how to process information from data and explicit problem statements. Their success mirrors many brain functions, and they enable the observed convergence across varied model designs.