Howdy! I’m not the most experienced on this topic or anything, but I’ve been playing around with ChatGPT for awhile now, and was interested in maybe providing a suggestion that could potentially help OpenAI in the long run, and ChatGPT indirectly. This is less a technical post, and more a business management suggestion. So, I’m not sure if this is the best place for it, but was really hoping someone from OpenAI would read and maybe pass it up.
The ability to build specific GPTs could definitely use with some quality of life updates, but if just a tiny bit more support went into fostering a creative ecosystem, it could very well give OpenAI a major competitive advantage. In my opinion, with threads from fully open-sourced models, and competition from other big models - the first of the big major models to pivot to a community-oriented approach is going to win.
I’ve always pointed to Valve as THE example of how to use community and good will to build a healthy ecosystem. They are not a nonprofit either, but they still understand that you can make decisions that benefit the interests of the company and the users, that creating a healthy ecosystem is far more profitable than simply creating fast high quality products, and giving experts time to play around is where you get real innovation. I’m not suggesting Open AI change to a flat organization structure, or even change their process for building new models - rather, to spend a few months finding ways users can help increase the value of what exists right now. If the last couple of years continue, OpenAI will always be scrambling to have the best model, and having to do typical marketing regardless of whether they do at that specific time. However, there’s a better way!
- GPT Builder 2.0
The interface could be much more helpful. Wishing I could sort the GPTs by the time they were developed was my inspiration for writing this post, but then I thought how information could be given for which model the GPT was optimized for, a better and more visible rating system could be created, user tags like Steam, etc. Further, the more tools that are given, the more the ecosystem could develop on its own. You’re never going to get a better deal than people working to enhance your product for free! You can reward people with things other than money like recognition and other perks like early access. Heck, Reddit Karma is fairly sought after and has no actual value.
- Listening to creatives and also reframing “AI slop” as a replacement creative commons
Actively promote GPTS and other community tools that enhance and highlight artists, voice actors, and more! Seeks out people from these groups and ask them what kind of tools would get them to be accepting of AI in their workflows. There are a lot of services that already do allow users to rate the best content and let others use it. Make this accessible and small part of a whole AI content ecosystem that is available for users. I can easily see a future where you ask ChatGPT to create an image. The user doesn’t like the image for their purpose but feels someone else might like it and then shares it with the community. The community loves it, upvotes it, downloads it, and some karma-like system is given to the user. Maybe even tip them, and/or pay for additional months of service.
The Karam-like system could be useless, or be traded for additional tokens - something that theoretically would encourage the user to create more cool stuff for the community. These incentives would also likely translate into better models as people would have better reason to prompt for quality over quantity. The only additional cost really is the storage but having access to nearly unlimited images, short videos, sounds, game assets, textures, writings like manuals but the value is simply in making them easy accessible and easy to integrate into people’s bigger projects.
- Customizability and integration of the AI ecosystem
Nobody is going to want to be forced into participating in this system which is why many of the services where you generate an image and then they force you to pay to keep it from being available to everyone is the wrong approach. Similarly, users should always be able to opt out - especially if future versions of ChatGPT are able to use knowledge of the marketplace itself to make suggestions based on these services and items. And even more especially, if the ecosystem develops into allowing paid GPTS. Nobody wants an Amazon-like situation where they constantly recommend you products.
- Free vs buyable GPTs
I think that is the biggest worry about paid GPTs, but allowing people to create more elaborate tools to enhance the base version and make money from it is almost always going to give you something better. Further, nothing stops OpenAI from buying up any really substantial GPT and making it available to everyone.
- Claude plays Pokemon
My background is in political psychology, not marketing, but I think projects like Claude Plays Pokemon on Twitch will be where actual performance is shown to the wider public, and maybe even more importantly where the next generation of kids will really learn how to get the most out of these models. OpenAI currently seems to be following a more Microsoft like approach and trying to integrate their service into as many businesses as possible. It’s a good strategy, but unlike operating systems, there’s always going to be more competition because you can’t prevent one model from being used to enhance others. Getting them early may give them an advantage for awhile, but businesses are cheap, and if they can do it on their own with more customizability, and locally - I just don’t think it’ll have enough pull on its own.
I don’t think it’s unreasonable to think you’ll have a cooperative AI + Twitch chat version in a year or two whereby the chat basically has to persuade the model to make the best moves. That’s just one of the way these models and groups might work together in solving games or puzzles or more. Even though it was taking forever, it was pretty fun to root for Claude and celebrate its victory.
I am now absolutely convinced that OpenAI should solicit requests for streamers wanting to do a few of these interactive projects and give those with the best proposals the resources and public attention - even if the model flops, people still had fun and will associate your model with that experience. Further, this could easily be a way to keep companies using OpenAI long-term. If they sponsor these types of events/games/projects, they get the good publicity and become associated with OpenAI: they may even get their own products featured (Minecraft seems like a great candidate) AND OpenAI doesn’t have to do much other than facilitate the process.
Heck, internally, if OpenAI wants to know something, they can literally out source the question as a competition and see what strategies people develop for free! In fact, presenting it as people solving a problem, and turning the solution effectively into a sporting event could be an entirely new industry and genre for streamers.
- Risqué content
I was hesitant to mention this one for self-evident reasons. However, it feels like a natural continuation of branding from the previous suggestion. I’ve been on the internet a long time, and so I’ve seen how a lot of these platforms have tackled mature content. Open AI should absolutely be concerned about alienating businesses for allowing mature content. That said, if they want to really make some money in the near future, some strategic branding could allow them to have their cake and eat it too. In fact, they could create a safer product and more safe for work experience by wading slowly into it.
My suggestion would be to change NOTHING with ChatGPT. However, having a subsidiary company that is designed at the start mostly to experiment with the boundaries of mature content like assisting people with writing more mature content, and chat bots, could make ChatGPT safer by being more willing to ask users to move to the subsidiary company after its given enough reason to believe its better suited for something like ChatAD (AfterDark). This means those people will keep the association with the ChatGPT brand and come back for other things, but major companies wouldn’t have to fear being associated with ChatAD. Lastly, since that type of content is part of life, I can’t imagine the information gained couldn’t be used in reverse for detecting more subtle bypassing of restrictions, or be used to create more restricted models for highly corporate environments.
Eventually, if those same open and community oriented principles are applied, OpenAI could end up helping sex workers stay safe while continuing to support themselves, and helping to reduce trafficking and fight more malicious “deep fakes”; by providing a substantial alternative, they’d reduce the number of people who feel inclined to go fully to open-sourced models which would provide much less accountability.
I know this is a fairy lengthy post and I’m not 100% sure it’s even in the best place, but hoping someone gives it a read. I did ask ChatGPT for a TL;DR:
TL;DR of Your Proposal:
You propose that OpenAI should pivot towards a community-driven approach to stay competitive against open-source and major AI competitors. Drawing inspiration from Valve’s ecosystem, you suggest that fostering user contributions and enhancing existing tools can be more beneficial than just focusing on building better models.
Key Suggestions:
- Improve GPT Builder (GPT Builder 2.0)
- Add sorting options, visible ratings, and tags for better navigation.
- Provide incentives like recognition, early access, or a karma system to encourage contributions.
- Promote Creative Users & AI-Generated Content
- Actively support artists, voice actors, and creatives by tailoring tools for them.
- Allow users to share AI-generated content (images, sounds, game assets, etc.), creating a Creative Commons-style AI repository.
- Introduce a karma-like system that rewards quality contributions and could be converted into service perks or tokens.
- Customizability & Opt-Out Options
- Users should not be forced into AI-generated marketplaces or recommendation systems.
- Avoid an Amazon-style aggressive recommendation approach to maintain user trust.
- Monetization of Paid GPTs
- Let users create and sell more advanced GPTs to encourage innovation.
- OpenAI could buy successful GPTs to make them publicly available.
- Public AI Demonstrations (e.g., AI Plays Pokémon on Twitch)
- Sponsor interactive AI projects to showcase model capabilities and engage the public.
- Encourage companies to integrate AI into real-world, interactive scenarios like gaming and problem-solving competitions.
- Crowdsourcing Innovation
- Host competitions where users develop strategies or solutions using AI, fostering engagement and innovation.
- Handling Mature Content (ChatAD/AfterDark Idea)
- Keep ChatGPT strictly SFW, but create a separate subsidiary (ChatAD) to explore mature content applications.
- This could help regulate content responsibly, keep major businesses comfortable, and prevent users from turning to riskier open-source alternatives.
- It could also help with safety issues like protecting sex workers and detecting harmful deepfakes.
Main Argument:
Rather than constantly racing to create the best AI model, OpenAI should invest in its community, incentivize high-quality contributions, and integrate AI into real-world applications through interactive content, creative tools, and community-driven projects.