Does Plugin Rhyme with App Store?

ChatGPT Growth Potential & How it Compares With Past Technological Revolutions

Introduction

It’s only March 2023, 4 months after the release of ChatGPT, and new advancements are being made every day. Only in the last 7 days, we’ve gotten: Databrick’s Dolly, GPT4All, the huge news of the open letter to hold AI advancement and so much more.

So much news!

One of the most recent and exciting developments is nonetheless still the introduction of ChatGPT plugins. Although still in an early Alpha guarded by a long waitlist, ChatGPT plugins are poised to usher in a new era of seamless integration between AI and third-party applications, much like the early days of the App Store. However, there’s a key difference: the barriers to entry for ChatGPT plugins are significantly lower, paving the way for rapid adoption and growth in the AI space.

These plugins have the potential to revolutionize how we interact with AI, making it more accessible, efficient, and useful than ever before. In this article, we’ll draw comparisons between the early days of the App Store and the adoption of ChatGPT plugins, highlighting the lower barriers to entry and the potential for rapid growth.

Excuse me, what Plugins?

Before we dive in, let’s quickly refresh our understanding of plugins. ChatGPT plugins, as described in OpenAI’s documentation, are third-party applications that enhance ChatGPT’s capabilities by connecting it to various APIs. This allows ChatGPT to retrieve real-time information, access knowledge bases, and even perform actions on behalf of users, such as booking flights or ordering food.

Very simplified visualization of where Plugins come in

What Rhymes with App Store?

When the App Store was first launched, it changed the way we interacted with our devices and opened up a world of possibilities for developers. However, the barriers to entry were quite high. Users needed to invest in expensive devices like iPhones and iPads, and developers had to learn new programming languages and syntax to build, test, and deploy apps.

The internet is abuzz with how this is the iPhone moment for A.I.–it is only natural, we are creatures of habit–we always try and use past data to predict the future. But empiric observation may not be enough when evaluating the possible impact of a game-changing technological revolution. It is akin to those that would’ve compared the internet with the invention of the Steam Engine. Yes, there are similitudes, but the pace of adoption was orders of magnitude faster.

From the time the engine was invented until an average Peter in London could jump on the first demo, it was 4 years, and another 17 years until the first “consumer-ready” product. This is of course because as technology advances, improvements are easy to realize and because software can move orders of magnitude faster than hardware.

Pure Software: Taking down the barriers

Thesis: In the case of the App Store, users needed to invest in expensive hardware (iPhones, iPads, etc.) to access apps. For ChatGPT plugins, users can access these AI-enhanced tools simply by opening a tab on their internet browser. This makes it significantly easier for users to try out and adopt these plugins without making a substantial investment upfront.

The potential for rapid adoption and growth: Empirical take

Below is a table showcasing the user adoption of the App Store over the 5 years after its launch in 2008.

As evident in the table, the user adoption of the App Store has grown exponentially since its launch in 2008, with a steady increase in both the number of users and cumulative app downloads. This data highlights the immense potential for AI-enhanced plugins like ChatGPT to experience similar growth trajectories.

The App Store though still has a physical limitation. Users needed access to both an expensive device (iPhone or iPad) and a network connection to get any app into their phone. Still, even with this limitation, its usage exploded, getting to its first 100 Million users in 4 years. Now, let’s take a look at the adoption rates of messaging apps WhatsApp, Line, and WeChat, measured by the years since their respective launches:

That looks even juicier, doesn’t it? 100 Mill in the first 3 years for WhatsApp and Line, 2 years for WeChat (not fair to compare, I know). Now let’s take a look at the little data we have from OpenAI’s ChatGPT:

That is something, isn’t it? Let’s see it all in perspective:

Ay Mamá!

How ChatGPT can outgrow our models: Elucubration

ChatGPT plugins, on the other hand, have significantly lower barriers to entry. Users can access these plugins by simply opening a tab on their internet browser, and developers can build them using the familiar OpenAPI format. This makes creating ChatGPT plugins easier and more accessible than developing traditional apps, which could lead to higher adoption rates.

The lower entry cost for ChatGPT plugins is a major contributing factor to their potential for rapid adoption. More people are likely to try out these AI-enhanced plugins without having to make a significant upfront investment. The ease of use, effectiveness, and security of these plugins will play a crucial role in their adoption and retention, and companies like OpenAI need to maintain credibility as secure data hosts.

So, it’s easier to access, easier to develop, and can have more impact. Let’s explore the potential directions this new market could take and how it will shape the future of AI-enhanced technology.

ChatGPT Plugins: Adoption Limits

Three key factors that will create an upper asymptote in the adoption of ChatGPT plugins are (1) security and trust in OpenAI, (2) Ease of user interaction, in this case, limited by text; and (3) the capacity of companies and developers to develop the interactions. Let’s explore these factors and how they will impact the growth of the plugin ecosystem.

(1) Security and trust in OpenAI

As the number of ChatGPT plugins grows, security and trust become increasingly important. With official plugins like Instacart, Kayak, and OpenTable, as well as numerous third-party plugins, users will be relying on these services to handle sensitive data and perform critical tasks.

To ensure the success of the ChatGPT plugin ecosystem, OpenAI must maintain its credibility as a secure platform that can be trusted to host users’ data securely. By fostering a strong sense of security and trust, OpenAI will encourage more users to try out these AI-enhanced plugins and contribute to their rapid adoption.

(2) User Interaction: Text UI & Beyond

While there might be an initial learning curve for some users who are accustomed to traditional UI, the majority of internet users are already familiar with chat interfaces from platforms like Messenger, WhatsApp, and Instagram. Businesses have been building bots and delivering complex interactions on these platforms for years, particularly in the APAC and LATAM regions, where Line/WeChat or WhatsApp interactions respectively between users and businesses are very much the norms.

Many large companies have recognized the potential of messaging apps for complex text interactions with their users. Here are just two notable examples:

1. KLM Royal Dutch Airlines — KLM uses Facebook Messenger, WhatsApp, and WeChat to offer a range of services to its customers, such as flight status updates, booking confirmations, and boarding passes. Additionally, KLM customers can use these messaging platforms to connect with customer service representatives for assistance with their travel-related queries.
2. Bank of America — developed its own AI-driven virtual financial assistant, Erica, which is integrated into the bank’s mobile app. Erica enables users to interact via text and voice, providing them with relevant financial information, helping them with transactions, and offering personalized financial advice.

These projects were, of course, much larger in scope than it would be to build a Plugin, but it showcases both the willingness of businesses to build text-based interfaces and the ability of users to interact with them to trigger actions in the real world.

Beyond text, but still GPT

Although the service is not publicly available, we know as well that GPT-4 has multimodal capabilities (i.e. can take images as input). This, combined with some prompt magic will be able to create interfaces that keep us where we are the most comfortable, whilst still being able to query models. It may get to be able to understand our screen and guide us without even direct text input.

Finally, an interpretation layer can do wonders. You can instruct a model to respond in a formatted way, where your code will interpret and render as defined by your rules. For an example of this in the wild, you can see this GPT3/3.5 example built by Shubham Verma, which takes text as input, and return a hierarchical graph.

(3) The Developer's Experience

Not only is it easier for users to access ChatGPT plugins, but it’s also less challenging for developers to create them. Developing a plugin primarily involves using the OpenAPI format, which is simpler than learning a new programming language or syntax (as was required to launch an App in the App Store). As these plugins focus on text interactions, developers can create impactful products without having to specialize or fill in multiple disciplines, such as UI/UX design. Allowing smaller teams taking on these projects, or companies to easily outsource their development to 3rd parties.

This lower entry cost is likely to contribute to a higher adoption rate, enabling a wider range of developers to enter the market and create innovative solutions. We are already seeing an explosion of Plugins available, and that is taking into account that OpenAI seems to be going very slowly down the Waitlist. (More on these new plugins below)

Future Outlook and Market Potential

As ChatGPT plugins gain traction, there are both exciting opportunities and potential challenges ahead. To fully realize the AI-enhanced interactions that ChatGPT plugins can offer we will have to wait for OpenAI to relax access to it. But let’s discuss some obstacles that may impact widespread adoption.

Opportunities in AI-enhanced interactions

ChatGPT plugins have the potential to revolutionize the way we interact with digital services. Some standout examples of plugins available today include:

• Speak: Learn how to say anything in another language with your AI-powered language tutor.
• Wolfram: Access computation, math, curated knowledge, and real-time data through Wolfram|Alpha and Wolfram Language.
• Instacart: Order from your favorite local grocery stores.
• Kayak: Search for flights, stays, and rental cars, and get personalized recommendations.
• OpenTable: Receive restaurant recommendations with a direct link to the book.

These plugins showcase a wide range of applications, from booking flights and ordering groceries to language learning and accessing complex computational knowledge. As more developers join the ecosystem, we can expect to see even more innovative use cases emerge, pushing the boundaries of AI-enhanced interactions.

Yet, many more use cases still are unexplored and we can see developers publishing new test plugins continuously trying to get ahead in this crazy market. Take a look at some of those you can find if you search today:

• Gerev AI: Search engine for your organization.
• Chat-todo-plugin: ChatGPT Plugin for managing a TODO list.
• ChadCode: An intelligence layer on your codebase.
• Domain Ideas: Brainstorm domain names with ChatGPT.
• Paperplane: Update SalesForce automatically.

Still, we can count the official plugins with the fingers on a Midjourney hand. This is an artificial limitation only created by OpenAI. And the reason is…

Challenges and obstacles to widespread adoption

Despite the immense potential of ChatGPT plugins, some challenges must be addressed to ensure widespread adoption:

1. Security and privacy: As we mentioned in Section III, OpenAI needs to maintain a strong security posture and user trust to ensure the success of the ChatGPT plugin ecosystem. This includes protecting user data and ensuring that third-party plugins adhere to stringent security standards.
2. Regulation and compliance: As the ChatGPT plugin ecosystem expands, it may become subject to increased regulatory scrutiny. OpenAI and plugin developers must navigate the complex landscape of global regulations, ensuring that their products are compliant with data protection, privacy, and other relevant laws.
3. Quality control: With a growing number of third-party plugins entering the market, maintaining a high standard of quality can become challenging. OpenAI must establish a robust review process for plugins and provide guidelines to developers that ensure a consistent level of quality across the ecosystem.

In conclusion, ChatGPT plugins have the potential to reshape the way we interact with AI and access digital services. By capitalizing on the opportunities offered by AI-enhanced interactions and addressing the challenges ahead, the ChatGPT plugin ecosystem can unlock new possibilities and drive innovation in the AI space.

A Bright Future with an ominous cloud

The exclusive dependency on software, standard formats used as well as the familiarity of chat UIs, combined with the new capabilities of ChatGPT plugins, creates a perfect environment for rapid adoption.

As more developers build plugins that cater to a wide range of use cases, users will become increasingly comfortable with incorporating ChatGPT plugins into their daily routines. This will help drive the growth and expansion of the ChatGPT plugin ecosystem, making it an essential part of how we interact with AI and access digital services tomorrow.

If only access to developing Plugins and working with APIs was equalized, we could have a tipping point that will drive users to drop google, Siri, etc in just a few months.

Get ready to witness a transformative shift in the way we interact with technology!