Get All Access for $5/mo

An OpenAI Rival Developed a Model That Appears to Have 'Metacognition,' Something Never Seen Before Publicly Anthropic's Claude 3 Opus did something never before seen from an AI model in internal tests: It recognized when a piece of its data seemed out of place and hypothesized that the detail was either a joke or a test.

By Sherin Shibu Edited by Melissa Malamut

Key Takeaways

  • Anthropic is the first to publicly speak about this particular kind of AI capability in internal tests.
  • Users on social media found the news "terrifying."
  • The company reportedly tried to cut hallucinations, or incorrect or misleading results, in half with its latest Claude rollout and inspire user trust by having AI tools cite sources.

A developer at Anthropic, an OpenAI rival reportedly in talks to raise $750 million in funding, revealed this week that its latest AI model appears to recognize when it is being tested.

The capability, which has never been seen before publicly, sparked a conversation about "metacognition" in AI or the potential for AI to monitor what it is doing and one day even self-correct.

Anthropic announced three new models: Claude 3 Sonnet and Claude 3 Opus, which are available to use now in 159 countries, and Claude 3 Haiku, which will be "available soon." The Opus model, which packs in the most powerful performance of the three, was the one that appeared to display a type of metacognition in internal tests, according to Anthropic prompt engineer Alex Albert.

"Fun story from our internal testing on Claude 3 Opus," Albert wrote on X, formerly Twitter. "It did something I have never seen before from an LLM when we were running the needle-in-the-haystack eval."

The evaluation involves placing a sentence (the "needle') into the "haystack" of a wider range of random documents and asking the AI about information contained only in the needle sentence.

"When we ran this test on Opus, we noticed some interesting behavior - it seemed to suspect that we were running an eval on it," Albert wrote.

According to Albert, Opus went beyond what the test was asking for by noticing that the needle sentence looked remarkably different from the rest of the documents. The AI was able to hypothesize that the researchers were conducting a test or that the fact the researcher asked for might, in fact, be a joke.

Related: JPMorgan Says Its AI Cash Flow Software Cut Human Work By Almost 90%

"This level of meta-awareness was very cool to see," Albert wrote.

Users on X had mixed feelings about Albert's post, with American psychologist Geoffrey Miller writing, "That fine line between 'fun story' and 'existentially terrifying horrorshow.'"

AI researcher Margaret Mitchell wrote: "That's fairly terrifying, no?"

Anthropic is the first to publicly speak about this particular kind of AI capability in internal tests.

According to Bloomberg, the company tried to cut hallucinations, or incorrect or misleading results, in half with its latest Claude rollout and inspire user trust by having the AI cite its sources.

Anthropic stated that Claude Opus "outperforms its peers" when compared to OpenAI's GPT-4 and GPT-3.5 and Google's Gemini 1.0 Ultra and 1.0 Pro. According to Anthropic, Opus shows "near-human" levels of understanding and fluency on tasks like solving math problems and reasoning on a graduate-school level.

Related: An AI Scam Stole 3 Million Site Visitors. Business Clones Are Pirating Services. Here's How to Prep Yourself for Alarming Trends in AI.

Google made similar comparisons when it launched Gemini in December, placing the Gemini Ultra alongside OpenAI's GPT-4 and showing that the Ultra's performance surpassed GPT-4's results on 30 of 32 academic benchmark tests.

"With a score of 90.0%, Gemini Ultra is the first model to outperform human experts on MMLU (massive multitask language understanding), which uses a combination of 57 subjects such as math, physics, history, law, medicine and ethics for testing both world knowledge and problem-solving abilities," Google stated in a blog post.

Sherin Shibu

Entrepreneur Staff

News Reporter

Sherin Shibu is a business news reporter at Entrepreneur.com. She previously worked for PCMag, Business Insider, The Messenger, and ZDNET as a reporter and copyeditor. Her areas of coverage encompass tech, business, strategy, finance, and even space. She is a Columbia University graduate.

Want to be an Entrepreneur Leadership Network contributor? Apply now to join.

Editor's Pick

Science & Technology

5 Rule-Bending AI Hacks to Make Your Mornings More Productive and Profitable

By 2025, AI will transform productivity by streamlining workflows and cutting costs. Major companies like Microsoft, Google, and OpenAI are leading the way, advancing AI into "Phase 3," where tools act as digital assistants. Discover 5 AI hacks to boost efficiency and redefine your daily routine.

Side Hustle

'Hustling Every Day': These Friends Started a Side Hustle With $2,500 Each — It 'Snowballed' to Over $500,000 and Became a Multimillion-Dollar Brand

Paris Emily Nicholson and Saskia Teje Jenkins had a 2020 brainstorm session that led to a lucrative business.

Marketing

5 Critical Mistakes to Avoid When Giving a Presentation

Are you tired of enduring dull presentations? Over the years, I have compiled a list of common presentation mistakes and how to avoid them. Here are my top five tips.

Business Ideas

63 Small Business Ideas to Start in 2024

We put together a list of the best, most profitable small business ideas for entrepreneurs to pursue in 2024.