A New Scale for AI Labs: Are They Aiming to Make Money or Just Research?

San Francisco: AI companies are at a special point. Many skilled workers from big tech are starting their own labs. Some researchers are focused on cool ideas rather than making money. Now, it’s hard to know who is really trying to earn cash.

To clear this up, I suggest a new scale for labs working on foundation models. This five-level scale measures how ambitious they are. It doesn’t focus on whether they are making money but rather if they want to.

Here’s the scale:

• Level 5: We are already making millions every day.

• Level 4: We have a plan to become the richest people on Earth.

• Level 3: We have exciting ideas that we will reveal later.

• Level 2: We have some ideas but no clear plan yet.

• Level 1: True wealth means loving yourself.

Major names like OpenAI are at Level 5. It gets interesting with new labs starting now, as their goals can be hard to decipher.

Many involved can pick their level freely. There’s so much cash in AI that no one will ask them for a detailed business plan. Even if a lab is just doing research, investors are still eager to join in. If a lab member isn’t driven to become a billionaire, they might prefer Level 2 over Level 5.

The challenge is that it’s not always clear which level an AI lab fits into. Much of the current buzz in the AI scene comes from this confusion. OpenAI’s shift from a non-profit to a for-profit caused a lot of worry, as they jumped from Level 1 to Level 5 quickly. On the flip side, Meta’s early AI work seemed like Level 2 when they really aimed for Level 4.

Let’s examine four major AI labs and where they stand on this scale:

Humans&
This lab made big news recently. Its founders have an exciting vision for AI models. They want to create tools to help communication and teamwork. However, they haven’t shared how these ideas will turn into products, so they seem to be at Level 3.

Thinking Machines Lab
This lab is tricky to rate! They have a former big shot from ChatGPT and raised $2 billion. So, I initially thought they were at Level 4. But after recent changes, including the CTO leaving, it seems there may not be a solid roadmap. They might be at Level 2 or 3 now.

World Labs
Fei-Fei Li is a respected name in AI. She raised $230 million for World Labs, focusing on AI that creates environments. Despite starting at Level 2, they have successfully launched products that meet industry needs, making them look like a Level 4 lab.

Safe Superintelligence (SSI)
Founded by Ilya Sutskever from OpenAI, SSI appears to be at Level 1. Sutskever keeps the lab away from commercial pressure, focusing on science instead. He turned down a buyout from Meta when raising $3 billion. However, if the research goes well, SSI could rise on the scale quickly.