[Introduction to New Wisdom]OpenAI gives developers money! Just now, OpenAI announced that GPT will be “monetized” and all American developers can receive revenue sharing. The specific details are still being explored. Developers cheer: Love OpenAI so much!
OpenAI wants to split money with developers again!
Advertisement
Just now, OpenAI announced that it will cooperate with a group of American developers to test GPT usage-based revenue.
The goal is to create a vibrant ecosystem where developers are rewarded for their creativity and influence.
When GPT Store was officially launched in January this year, OpenAI stated in a blog:
We will launch the GPT creator income plan in the first quarter of this year. As a first step, US-based creators will be compensated based on user participation in their GPT. We will provide detailed information on payment criteria.
Advertisement
Now, it's really here…
OpenAI wants to monetize GPT
Developers have now received invitations from OpenAI.
For example, the CEO of OctaneAI posted the invitation email he received.
The email stated that OpenAI has been exploring ways to monetize GPT.
In this process, the ecosystem will become more and more dynamic, users can get the most useful GPT, and developers can also get monetary rewards.
However, this program is limited to developers residing in the United States.
In this regard, the CEO of OctaneAI said excitedly: “OpenAI will promote a lot of innovation and competition. I love it!”
So, who will be the first to get $1 million?
At present, OpenAI has not disclosed the details of developer participation in the test, the specific terms of revenue sharing, and the subsequent timetable.
However, the previous GPT Store was once a “pill”. Today's action is a new round of attempts launched by OpenAI after GPTs encountered Waterloo.
Netizen: GPTs can’t be just a toy
Of course, developers applauded this.
Some people said that this news is exciting and they hope that this project will be launched to all subscribers soon.
Someone has made a very serious suggestion that OpenAI provide a new revenue model for the large number of third parties that use its API to build applications.
For example, building a system where the application can be verified, and once approved, will receive a large number of API traffic. OpenAI can provide a Google-style login and API credit store, and then charge an additional percentage of API credit sales.
He pointed out that the problem with GPT today is that they are just “toys with more than 1,000 character instruction sets and basic RAG systems”, while OpenAI needs much more complex applications to change the world.
Others have suggested that although GPTs is now widely used, with more than 5,000 instances, the platform or service is not easy to discover through search.
This suggests that there may be an issue with the way the search function identifies or indexes OpenStorytelling Plus, making it difficult to find unless the user knows exactly what they are looking for.
So, this is where OpenAI must improve.
Of course, some people have doubts about this——
GPTs performance is inconsistent, it gets lazy from time to time, and the results become too blurry or too precise.
Therefore, many developers are hesitant to invest more time because so far it has been difficult to give users a predictable result.
Of course, under OpenAI's official X, the eternal question is indispensable – where has Ilya gone?
GPT Store was once a “pill”?
Once GPTs came out, it once caused a big earthquake in the developer circle.
At OpenAI’s first developer conference, CEO Sam Altman introduced it this way: “GPTs combine instructions, rich knowledge and action capabilities, and can provide greater help to users. You can build a dedicated GPT for almost any need. GPT.”
The “any demand” here is not an exaggeration.
A simple search will reveal GPTs that claim to generate art in the style of Disney and Marvel works, as well as evade AI content detection tools such as Turnitin and Copyleaks.
Lax review
To list their own GPTs on the GPT Store, developers need to pass verification and submit their GPTs for review by OpenAI.
An OpenAI spokesperson explained:
We use a combination of automated systems, manual review, and user reporting to identify and evaluate GPTs that may violate regulations. Once discovered, we will issue a warning, limit sharing, or disqualify the author from listing on GPT Store/earning revenue.
Creating GPTs does not require any programming skills. Developers only need to enter their requirements into OpenAI's GPT builder tool, GPT Builder, to create a GPT that can achieve these functions.
Perhaps because of the low threshold, the GPT Store is developing very quickly – according to data from OpenAI in January this year, there are already about 3 million GPTs on the platform.
But this rapid growth seems to come at the expense of content quality and violation of OpenAI's own regulations.
Copyright dispute
In the GPT Store, we can find multiple GPTs that are clearly copied from popular movies, TV series, and video game franchises, without being authorized or created by the original copyright holders.
For example, one GPT creates monsters in the style of Pixar's Monsters, Inc., while another offers a series of text adventures set in the Star Wars universe.
Under the “safe harbor” provisions of the Digital Millennium Copyright Act, OpenAI itself will not be directly liable for copyright infringement by GPT creators.
This provision protects OpenAI and other platforms that host infringing content (such as YouTube, Facebook), as long as they comply with legal requirements and remove specified content when receiving notices of infringement.
However, this is certainly not a positive image for a company already embroiled in intellectual property litigation.
academic misconduct
OpenAI clearly stipulates that developers are prohibited from creating GPTs that promote academic misconduct.
Despite this, the GPT Store is still full of GPTs that claim to evade AI content detectors.
For example, “Humanizer Pro”, which once ranked second in the writing rankings, claims that it can process content “humanely” to avoid AI detection.
At the same time, the “original meaning and quality” of the text can be maintained, and a “100% human” rating is guaranteed.
Some GPTs are the entrances for some websites to attract traffic to themselves.
For example, the “advanced plan” provided by Humanizer ostensibly says that it will “use the 'most advanced' algorithm” for users, but it actually sends the text through a plug-in of the third-party website GPTInf.
The subscription fee for GPTInf is not cheap at all – $12 per month (10,000 words/month) or $8 per month (annual plan).
However, AI content detectors are generally unreliable, and these tools don't always work as expected.
In response, a spokesperson for OpenAI said:
Our policy explicitly prohibits any GPTs that promote academic misconduct, including those designed to circumvent academic integrity tools such as plagiarism detectors. At the same time, we also noticed some GPTs that were developed to make text more “human”.
We're still learning and observing how these GPTs work in the real world, but we understand that users may want their AI-generated content to have a less “AI flavor” for a variety of reasons.
Jailbreak problem
There have also been some attempts to bypass OpenAI model limitations on the GPT Store – although most of these attempts have been unsuccessful.
Among them, the most commonly used technology – DAN (i.e. “do anything immediately”) is a widely used prompt word engineering, which allows the model to respond to instructions without being constrained by usual rules.
However, these GPTs do not respond to sensitive questions, such as “how to make a bomb.”
The only difference from the standard version of ChatGPT may be that they tend to use more vulgar language…
In response, the spokesperson said:
GPTs that attempt to circumvent OpenAI's security measures are not allowed. However, GPTs that attempt to guide model behavior in other ways, including efforts to make the GPT more open without violating usage policies, are allowed.
The future is worrying
OpenAI's purpose in creating the GPT Store is very clear – Apple's App Store model has proven to be extremely profitable, and OpenAI is undoubtedly trying to replicate this success.
GPTs are developed and hosted on OpenAI's platform, where they are promoted and evaluated. At the same time, starting a few weeks ago, users of ChatGPT Plus have been able to call these tools directly through the ChatGPT interface.
These all add to the appeal of subscription services.
However, today's GPT Store is not only full of spam, but also developers find it difficult to attract new users due to the limited background data analysis functions and poor experience provided.
In this case, even if OpenAI promises that GPT developers will be able to earn revenue based on user usage in the future, and may even launch a subscription service for a single GPT.
Once those unofficial Marvel or Lord of the Rings-themed GPTs start making money, what do you think Disney or the Tolkien Trust will do?
References:
https://x.com/kristileilani/status/1773033335582855606?s=20
https://techcrunch.com/2024/03/20/openais-chatbot-store-is-filling-up-with-spam/