General Intuition Raises Major Funding to Improve AI World Models Using Gaming Data

General Intuition 獲得巨額融資，利用遊戲數據提升 AI 世界模型

Jun 25, 2026, 16:55

Introduction

General Intuition, a startup that focuses on how AI understands space and time, has announced a $320 million funding round to grow its agentic AI models.

專注於 AI 如何理解空間與時間的初創公司 General Intuition 宣布，已獲得 3.2 億美元融資，用於發展其 agentic AI 模型。

Main Body

The company, which started as a spin-off from the gaming platform Medal, uses a special dataset containing millions of hours of gameplay. Unlike other current methods that only look at images, General Intuition uses action labels to teach its models about cause-and-effect and how to move through spaces. This method helps the AI move from simulated environments—which they call "the gym"—into physical robots. For example, they have already successfully used this technology to train a four-legged robot using very little real-world data.

這家公司最初是由遊戲平台 Medal 分拆而成的，使用一個包含數百萬小時遊戲過程的特殊數據集。與目前其他僅分析圖像的方法不同，General Intuition 使用行動標記來教導其模型關於因果關係以及如何在空間中移動。這種方法幫助 AI 從模擬環境（他們稱之為「gym」）遷移到實體機器人上。例如，他們已經成功利用這項技術，在極少現實世界數據的情況下訓練出一個四足機器人。

This project has received significant financial support, with the latest funding round led by Khosla Ventures and other famous investors like Jeff Bezos and Eric Schmidt. This investment brings the company's total value to $2.3 billion. The money will be used mainly to increase their computing power through a partnership with CoreWeave and to release their API to more users. Vinod Khosla emphasized that the company's unique data is a key factor in developing AI intuition, making the firm a valuable long-term asset.

這個項目得到了顯著的財務支持，最新一輪融資由 Khosla Ventures 領投，其他知名投資者如 Jeff Bezos 和 Eric Schmidt 也參與其中。此次投資使公司的總價值達到 23 億美元。這筆資金將主要用於透過與 CoreWeave 合作來增加計算能力，並將其 API 開放給更多使用者。Vinod Khosla 強調，公司獨有的數據是開發 AI 直覺的關鍵因素，使該公司成為一個極具價值的長期資產。

Additionally, CEO Pim de Witte has established a strict set of ethical rules. The company has clearly forbidden the use of its technology for lethal autonomous weapons, although it allows the AI to be used for search and rescue missions. Furthermore, the company launched "Nerve," a marketplace that helps gamers earn money through data labeling and robot operation to prevent job losses caused by AI. The overall goal is to provide the basic technology that other developers can use for robotics and simulations.

此外，執行長 Pim de Witte 建立了一套嚴格的倫理準則。公司明確禁止將其技術用於致命自主武器，儘管允許 AI 用於搜索與救援任務。此外，公司推出了名為 "Nerve" 的市場，幫助遊戲玩家透過數據標記與操作機器人來獲利，以防止 AI 造成的失業問題。整體目標是提供基礎技術，供其他開發者用於機器人研發與模擬。

Conclusion

General Intuition is now expanding its computing systems and preparing to release its API to prove that its simulations can work effectively in the real world.

General Intuition 目前正在擴展其計算系統並準備發布 API，以證明其模擬系統在現實世界中能有效運作。

Vocabulary Learning

⚡ The 'Action-Result' Logic: Moving from A2 to B2

At the A2 level, we usually describe things using simple sentences: "The AI learns. It uses data." To reach B2, you need to connect actions to their purposes or consequences.

Look at this phrase from the text:

"...uses action labels to teach its models about cause-and-effect..."

🛠 The Power Move: "To + Verb" (Purpose)

Instead of saying "They have a tool. They want to help robots," a B2 speaker says: "They use action labels to teach its models."

How to apply this today: Stop using "because" for every reason. Start using [Action] + [to + Verb].

A2: I study English because I want a better job.
B2: I study English to find a better job.

🚀 Level Up: Complex Nouns (The 'B2 Glue')

Notice how the author doesn't just say "rules," but uses "a strict set of ethical rules."

To sound more fluent, don't just use a noun. Wrap it in a 'description layer':

Basic: Rules $\rightarrow$ B2: A strict set of ethical rules.
Basic: Funding $\rightarrow$ B2: A significant financial support.
Basic: Data $\rightarrow$ B2: A unique dataset.

🧠 Quick Concept: "Cause-and-Effect"

In the article, this is a key term. At B2, you stop describing what happened and start explaining why it happens.

Example from the text: The company uses gaming data $\rightarrow$ (Cause) $\rightarrow$ The AI understands space $\rightarrow$ (Effect).

When you speak, try to use the word "consequently" or the phrase "leads to" to bridge your ideas.

"General Intuition uses gaming data; consequently, their robots learn faster with less real-world data."

Vocabulary Learning

spin-off (n.)

A new company formed from a part of an existing larger company

Example:The new biotech firm is a spin-off from the university's research department.

simulated (adj.)

Created by a computer program to represent a real-life situation

Example:Pilots often train in simulated environments before flying a real aircraft.

emphasized (v.)

To give special importance or prominence to something in speaking or writing

Example:The manager emphasized the importance of meeting the deadline.

asset (n.)

A useful or valuable thing, person, or quality

Example:Her ability to speak three languages is a huge asset to the international team.

established (v.)

To set up a system, rule, or organization on a firm or permanent basis

Example:The company established a new set of guidelines for remote work.

forbidden (adj.)

Not allowed; banned by a rule or law

Example:Smoking is strictly forbidden inside the hospital building.

lethal (adj.)

Sufficient to cause death; deadly

Example:The soldiers were equipped with lethal weapons for defense.

autonomous (adj.)

Acting independently or having the freedom to do so; self-governing

Example:Many car manufacturers are developing fully autonomous vehicles.

Practice B2 words in a crossword→