With the introduction of large language models (LLM), machine learning (ML) and artificial intelligence (AI) are being used by everyday developers for the first time. These amazing applications, even software with billions in R&D spending that were previously almost impossible for even large tech companies to build, are suddenly not only possible, but able to be built and shared.
The upward trend of artificial intelligence creation starts in 2021, grows rapidly in 2022, and explodes in the first half of 2023. With the emergence of more and more LLM providers (such as Google, OpenAI, Cohere, Anthropic) and development tools (such as ChromaDB, LangChain), the development speed has increased. At the same time, the natural language interface of the generated code makes its creation more accessible to more people than ever before.
Here to share some statistics on the state of AI development.
artificial intelligence creation
Since Q4 2022, we have seen an explosion of AI projects. As of the end of 2Q23, there were almost 300,000 different projects related to AI. In comparison, a search of GitHub reveals only about 33k OpenAI repositories over the same time period.
About 160,000 of these items were created in 2Q23. That's about an 80% increase quarter-over-quarter and a 34-fold increase year-over-year. And continue to see those numbers accelerate.
Most of these projects are using OpenAI. When we compare providers, OpenAI dominates more than 80% of the AI projects in the market. The OpenAI GPT-3.5 Turbo template currently has more than 8000 forks. But there are signs that things may be changing.
In 2Q23, we saw:
OpenAI projects span +125k (80% increase)
Coherent projects crossed +1k (a 100% increase)
Anthropic and Google projects stay <1k
The emergence of LangChain
One of the biggest names in AI activity is LangChain. Using LangChain as a wrapper for some of these models has accelerated development and continues to see massive adoption.
As of 2Q23, there are nearly 25,000 active LangChain projects on Replit + 20,000 of which were created during that quarter, a 400% increase from the previous quarter.
It is important to note that LangChain provides enough abstraction around LLM providers to allow developers to switch easily. The development of this project may have played a role in the rise of new LLM providers and open-source LLMs.
Takeoff School, founded by Mckay Wrigley, has a course called LangChain 101, where people can start learning LangChain today. The project is about to pass 1000 forks.
The Rise of the Open Source Model
We're also seeing an increase in projects leveraging the open source model. Hugging Face and Replicate are two API providers and SDKs that are important entry points into the open source model.
In 2Q23, we had more than 5,000 projects using the open source model. The cumulative number increased by 141% year-on-year. More than 70% of projects utilize Hugging Face, but the usage of Replicate has increased by nearly 6 times QoQ.
Classification of programming languages
Interestingly, we see very similar growth rates for Python and JavaScript, Python being the slightly more common language in AI development. JavaScript, however, grew slightly faster in the second quarter.
It's worth noting that projects can have both Python and JavaScript. The two are not mutually exclusive. Many, if not most, projects have a Python backend and a JavaScript frontend.
Interestingly, language varies by geography. Some locales use JavaScript instead of Python.
In the past 90 days, about 50,000 Python developers:
United States: 32%
India: 11%
UK: 7%
Canada: 3%
Brazil: 3%
However JavaScript looks very different. Of the approximately 34,000 JavaScript developers, the percentages are by region:
United States: 22%
Indonesia: 10%
India: 9%
Vietnam: 7%
Philippines: 5%
JavaScript developers in AI tend to be more Asian, while Python developers are more North American. India appears relatively evenly represented.
what are they building
The apps being built are fantastic, and we can't do justice to all of them here. It’s super exciting that artificial intelligence is paving the way for a new generation of entrepreneurs to build previously impossible applications. Examples include:
CampLingo: Generate language learning products.
NodePad.space - Visual ideas for artificial intelligence.
School of Magic - Artificial Intelligence Tools for Educators.
MightyDeals AI - Make Affordable Deals on the Internet (Story).
AI Avatars with LeapAPI - Templates for creating professional or themed avatars.
BabyAGI - An agent that can read and write its own code.
From supporting underserved situations, to now autonomous agents capable of taking their own actions, these projects are evolving every day, and very rapidly.