OpenAI has been making waves in the artificial intelligence space with their chatbot ChatGPT, which can hold natural conversations and generate human-like text on demand.
After launching ChatGPT to the public late 2022, OpenAI is now unveiling a series of major new capabilities and products to enhance the platform.
The ChatGPT Store Arrives
One of the most anticipated launches is the new ChatGPT Store, which provides access to a marketplace of customized chatbots called ChatGPTs (GPTs). This store is now rolling out to Plus subscribers, allowing them to browse specialized GPTs created by developers for different uses.
The storefront showcases trending and recommended GPTs each week, curated by OpenAI. These featured bots provide a way to discover some of the most popular and useful GPTs that the community has built. There are also categories to filter GPTs by topics like productivity, research, education, programming, and more.
Users can search the store by keywords to find GPTs for specific use cases, like "YouTube" or "finance." The store is populated with GPTs built by independent developers as well as bots created by OpenAI itself. Some examples include a video summarizer, SEO assistant, math solver, and meditation guide.
Creating and Monetizing Custom GPTs
In addition to browsing existing GPTs, the store also allows users to build and submit their own custom bots. There is a creation tool that guides users through training a new GPT with customized data and prompts. This system lowers the barrier to building specialized bots that tap into unique APIs and datasets.
OpenAI mentioned that in Q1 2023 they will launch a revenue-sharing program for GPT creators, initially paying based on user engagement. This monetization model provides an incentive for developers to build valuable GPTs that rise to the top in popularity. Details are still sparse, but it's a first step towards a sustainable marketplace.
Of course, there are challenges to profiting from custom GPTs. Since the knowledge is crowdsourced, popular innovations could get replicated by others. Unique integrations with proprietary data and APIs may be harder to mimic though, creating more defensible value. Overall, the opportunity exists for some standout GPTs to generate income through the engagement-based model.
New Team Plan Unlocks More Advanced Features
Alongside the store, OpenAI introduced ChatGPT Professional for teams. This plan provides access to capabilities beyond the standard Plus offering, including:
- A larger conversational context window of 32,000 tokens (vs. 4,000 for Plus)
- Higher daily message limits and compute power
- Advanced instructiontuningandchaining to make GPTs more useful
- Streamlined workflows for collaborating on ChatGPT conversations
The Team plan starts at $25 per user per month, billed annually for a minimum of two users. The expanded context window and computing resources enable more complex dialogues with increased accuracy and memory. Teams can also better customize ChatGPT's behavior for their unique needs.
While the standard Plus plan suffices for basic use cases, the Professional version unlocks more robust features suitable for enterprise use. Companies can implement it to boost productivity, conduct research, create content, answer customer questions, and automate processes. The shared team environment improves coordination and transparency.
Adding Memory and User Preferences
One limitation of the original ChatGPT model was its lack of persistent memory. Each conversation started from scratch, without retaining specifics from previous chats. This led to repetitive conversations and a need to re-explain context regularly.
Now there are signs that OpenAI is testing memory and user preference features that allow ChatGPT to learn interactively during conversations. Some accounts have reported seeing settings to toggle on memory, with the bot gaining context between chats to provide more personalized and relevant responses.
This advancement addresses a major shortcoming compared to human conversation. With memory, ChatGPT can become more helpful by recalling earlier details, conversations, and preferences over an extended dialogue. It can progressively improve through these interactions and accumulate useful knowledge unique to each user.
OpenAI seems to be rolling out this capability slowly and cautiously. Long-term memory raises moderation challenges, since the bot could potentially retain harmful, biased, or inconsistent information without careful design. But implemented responsibly, memory represents a big step forward in making conversations feel more natural.
Examining the Content Copyright Lawsuits
ChatGPT's rapid impact across education, business, and culture has not come without controversy. OpenAI currently faces lawsuits from literary organizations like the Authors Guild and media outlets like the New York Times alleging copyright infringement and citing instances of ChatGPT copying original written content verbatim.
In response, OpenAI asserts that the training data used was general domain public knowledge under fair use provisions. They claim verbatim regurgitation of copyrighted material is a rare bug, not intended behavior. The company suggests that any examples of direct copying were induced by manipulative prompts feeding the model lengthy excerpts.
According to OpenAI, their models don't typically exhibit this regurgitation capability without special prompting. They aim to reduce such vulnerabilities through technical fixes and policy changes. Overall, OpenAI believes they are operating legally and ethically in training generative AI on publicly available data. But they acknowledge improvements must be made.
It remains to be seen how the clash between AI capabilities and copyright law will unfold. Content creators want protection from automatic recreation of their work, while also benefiting from the technology's creative potential. New norms, protections, and business models may need to emerge to reach an equilibrium. But for now, the focus is on making models less prone to duplication regardless of prompts.
The Path Ahead: Steady Expansion Likely
With these myriad updates to ChatGPT plus continuing model training, OpenAI is showing their ambition to rapidly iterate and expand the platform's capabilities. More enhancements are likely on the horizon as the technology continues advancing.
We can expect a steady rollout of new features like memory, specialized skills, and speed/accuracy boosts through architecture improvements. OpenAI will also need to educate users on responsible AI practices as the technology grows more powerful.
On the product side, offerings tailored for education, finance, healthcare, and other industries could be in store. OpenAI will have to balance open access with generating revenue, possibly through subscriptions, enterprise services, and creator monetization programs.
There is massive interest in generative AI right now, with both hype and backlash. While progress will be uneven, ChatGPT seems poised for regular enhancement as a leader in the space. With each update, it takes a step closer to the ultimate vision of building human-like artificial general intelligence.