On March 4, ChatGPT, an intelligent chat tool launched by American AI research company OpenAI, has set off the so-called generative AI (AIGC) craze around the world, and it has been integrated into social, email, educational and many other applications. So who makes most of the money between OpenAI and its ChatGPT customers?

On March 4, ChatGPT, an intelligent chat tool launched by American AI research company OpenAI, has set off the so-called generative AI (AIGC) craze around the world, and it has been integrated into social, email, educational and many other applications. So who makes most of the money between OpenAI and its ChatGPT customers?

Notion is just one of the first companies to eat crabs in the current wave of racing to take advantage of the surge in interest in AIGC. This week, Snapchat launched an integrated ChatGPT service called My AI for users who pay $4 a month to subscribe to the Snapchat Plus service.

Educational application Quizlet has also released Q-Chat, a ChatGPT-based tutoring software. Instacart, the US fresh ecommerce company, said it was developing a tool that would allow customers to query food and get data about the product from its retail partners to make decisions about whether it was worth buying.

As more and more companies adopt ChatGPT, we can't help but wonder: who earns most of that revenue? Will more and more companies find ways to integrate AI into products that can be charged? Or does most of the profits end up going to a few companies that provide the basic models for other companies to build and refine their own AI tools?

The answer will largely determine whether AIGC represents a true platform shift from desktops to mobile phones or just a limited set of innovations whose benefits are ultimately concentrated in a handful of big winners.

Developers are obviously thinking about this, too. Recently, in response to concerns, OpenAI said it would no longer use developer data to improve its model without the developer's permission. Instead, it will require developers to choose whether or not to join.

Greg Brockmann, president and chairman of OpenAI, said: "one of our biggest concerns is to figure out how we can become super developer-friendly." Our mission is to really build a platform on which others can build business. "

Maybe it's as simple as that: developers don't want to help OpenAI improve its model for free, and OpenAI decides to respect their wishes. This explanation seems to be more in line with the idea that AI really represents platform transfer. Or, OpenAI believes that the model can continue to improve quickly, whether or not all of these developers choose to join. This explanation means that OpenAI and a few other companies will get most of the return.

So what kind of AI features are these companies actually selling? At present, the AI products on the market are basically white-label versions of ChatGPT. Starting this week, OpenAI will provide its technology to other companies through API. Any company can resell ChatGPT in its own application for only $0.002 per 1000 tokens.

For now, when it comes to AIGC, consumers don't have much choice. To some extent, the options in the interface seem to be richer. Do you want to draft an email in AI? It may be more convenient in Notion because you already have some minutes of the meeting. Do you want to solicit some recipe ideas or ask some questions? If your laptop is not with you, asking My AI on Snapchat may be the quickest way.

There are still billions of people who have never used ChatGPT. Introducing the feature to popular consumer apps like Snapchat, which has 750 million users a month, could help it find a new audience. Paying Notion or Snapchat for the use of such ChatGPT features also ensures user stickiness. Over time, these tools will become more personalized, as applications will use the data provided by their users to help improve the underlying model of OpenAI.

Zhao Yi, CEO of Notion, said that more features supported by ChatGPT are coming. At first, it may just be for writing and editing, but a deeper change is coming. "I've never been so excited about anything," he said. The AIGC represented by ChatGPT feels like electricity, while writing and editing are like light bulbs, which are the first use cases of electricity. However, there are many other household appliances that need electricity. "

Zhao Yi, CEO of Notion, said that more features supported by ChatGPT are coming. At first, it may just be for writing and editing, but a deeper change is coming. "I've never been so excited about anything," he said. The AIGC represented by ChatGPT feels like electricity, while writing and editing are like light bulbs, which are the first use cases of electricity. However, there are many other household appliances that need electricity. "

Zhao Yi, CEO of Notion, said that more features supported by ChatGPT are coming. At first, it may just be for writing and editing, but a deeper change is coming. "I've never been so excited about anything," he said. The AIGC represented by ChatGPT feels like electricity, while writing and editing are like light bulbs, which are the first use cases of electricity. However, there are many other household appliances that need electricity. "

Zhao Yi, CEO of Notion, said that more features supported by ChatGPT are coming. At first, it may just be for writing and editing, but a deeper change is coming. "I've never been so excited about anything," he said. The AIGC represented by ChatGPT feels like electricity, while writing and editing are like light bulbs, which are the first use cases of electricity. However, there are many other household appliances that need electricity. "

Perhaps the best-case scenario for small companies is that the AIGC industry becomes similar to the cloud computing market. The infrastructure will be built by a few companies, but the features they provide at low cost will inspire a whole new generation of startups.

At the same time, the capabilities of ChatGPT and other large language models will only expand, and they are likely to engulf many AI startups in the process. As technology companies integrate AI functionality into their applications this year, they'd better figure out the huge risks as soon as possible.