Skip to main content

CAPTCHAs may be a thing of the past, thanks to new machine learning research

CAPTCHA is an acronym for Completely Automated Public Turing test to tell Computers and Humans Apart. The term was coined in 2003, when the use of automated bots was becoming commonplace, and it refers to those annoying squiggly distorted letters that you have to type in when creating an online account. Although some companies have found ways around them, CAPTCHAs are still ubiquitous online.

Researchers at the AI company Vicarious may have just made CAPTCHAs obsolete, however, by creating a machine-learning algorithm that mimics the human brain. To simulate the human capacity for what is often described as “common sense,” the scientists built a computer vision model dubbed the Recursive Cortical Network.

Recommended Videos

“For common sense to be effective it needs to be amenable to answer a variety of hypotheticals — a faculty that we call imagination,” they noted in a post at their blog.

The ability to decipher CAPTCHAs has become something of a benchmark for artificial intelligence research. The new Vicarious model, published in the journal Science, cracks the fundamentals of the CATCHPA code by parsing the text using techniques that are derived from human reasoning. We can easily recognize the letter A for example, even if it’s partly obscured or turned upside down.

As Dileep George, the co-founder of the company explained to NPR, the RCN takes far less training and repetition to learn to recognize characters by building its own version of a neural network. “So if you expose it to As and Bs and different characters, it will build its own internal model of what those characters are supposed to look like,” he said. “So it would say, these are the contours of the letter, this is the interior of the letter, this is the background, etc.”

These various features get put into groups, creating a hierarchal “tree” of related features. After several passes, the data is given a score for evaluation. CAPTCHAs can be identified with a high degree of accuracy. The RCN was able to crack the BotDetect system with 57 percent accuracy with far less training than conventional “deep learning” algorithms, which rely more on brute force and require tens of thousands of images before they can understand CAPTCHAs with any degree of accuracy.

Solving CATCHPAs is not the goal of the research, but it provides insight into how our brains work and how computers can replicate it, NYU’s Brenden Lake told Axios. “It’s an application that not everybody needs,” he said. “Whereas object recognition is something that our minds do every second of every day.”

“Biology has put a scaffolding in our brain that is suitable for working with this world. It makes the brain learn quickly,” George said. “So we copy those insights from nature and put it in our model. Similar things can be done in neural networks.”

Mark Austin
Former Digital Trends Contributor
Mark’s first encounter with high-tech was a TRS-80. He spent 20 years working for Nintendo and Xbox as a writer and…
My dream AI MacBook may be delayed until 2027 thanks to Apple Intelligence
Apple's Craig Federighi discussing Apple Intelligence at the Worldwide Developers Conference (WWDC) 2024.

Artificial intelligence (AI) is experiencing explosive growth at the moment, with everyone in the tech world seemingly trying to get in on the action. That includes Apple, but it’s no secret that the company’s Apple Intelligence platform is struggling to compete with the likes of ChatGPT, Gemini and Copilot. Yet we’ve just had some news that could make that situation even worse, especially for Mac users.

That’s because Bloomberg reporter Mark Gurman has just claimed that some key Apple Intelligence features won’t be available until 2026 or even 2027, putting the dream of a powerful AI-powered MacBook firmly on the backburner.

Read more
Anthropic’s new Claude model offers both real-time and long-pondered responses
Claude code homescreen

OpenAI's o3 and DeepSeek's R1 models have some new competition. Anthropic announced Monday the release of its new "hybrid reasoning" model, Claude 3.7 Sonnet.

Existing reasoning models like o3, R1, and Google’s Gemini 2.0 Flash Thinking are designed to break down complex problems into smaller tasks, then deduce and verify their answers before responding, a process that returns more accurate answers at the cost of higher compute usage and longer inference times. Claude 3.7 Sonnet, on the other hand is capable of providing either "near-instant responses or extended, step-by-step thinking that is made visible to the user," according to the company's announcement post.

Read more
Meta’s new ‘Llamacon’ event is all about open-source AI
A silhouetted person holds a smartphone displaying the Facebook logo. They are standing in front of a sign showing the Meta logo.

Meta announced on Tuesday that it is launching a new developers conference in April, dubbed "Llamacon," that will focus on “open source AI developments.”

The event is scheduled to take place April 29, 2025 and comes on the heels of "the unprecedented growth and momentum of our open-source Llama collection of models and tools," in an announcement post. The company has not shared any additional details, such as where the conference will take place or how much ticket prices will run, but the company promises to share more details "in the coming weeks."

Read more