You’ve probably heard artificial intelligence models described as “open” or “closed.” These are not descriptions of the model’s personality. Large language model AIs like the one under the hood of ChatGPT, contrary to appearances, have no real personality.
The designations refer to whether all information about how an AI model works is publicly available and the model can be changed, or whether the model’s developer keeps its inner workings secret and the model itself is private property.
Table of Contents
Open source software
The concept of open source software emerged in the free software movement of the 1980s and 1990s. The movement’s founders believed that software developers and users had the right to “four freedoms” – running the program, studying and modifying it, distributing copies of the original, and distributing copies of later modified versions. The basic requirement was that the source code – the basic instructions – of a program should be made available.
In the late 1990s, software developers involved in projects such as the Netscape web browser and the Linux operating system coined and promoted the term “open source” to refer to these ideals.
As part of the evolving movement, certain organizations developed open source licenses that specified how a particular source code may be used and distributed, including the Gnu General Public License, the Apache License, the MIT License, and the Berkeley Software Distribution. Each type of license also specified any possible restrictions on the application of software patents to the source code.
Open Source or Open Weight?
The open source idea has regained traction in recent years as large artificial intelligence language models have gained traction, most notably OpenAI’s ChatGPT, released in 2022. Developers first train new models on large data sets and then make the models available for other people to use.
Open source artificial intelligence is explained in two minutes.
Meta was one of the first major companies to release an open source model for large languages called LLaMa. The company released LLaMa on February 24, 2023, providing the “inference” source code – the instructions that run the model. And it released the so-called weights, the encrypted knowledge that the model learned during training. However, open source organizations such as the Open Source Initiative have stated that the LLaMa licensing guidelines prohibit commercial reuse, with the initiative claiming that it is not truly open source.
Other companies have released “open weight” models, such as DeepSeek AI’s DeepSeek and Alibaba’s Qwen. The models have less restrictive conditions for reuse and were quickly adopted by the AI community. Still, many developers believe that a true open source AI model must include not only the source code and weights, but also the data used to train the model.
There is a lot to open
The Open Source Initiative’s definition of a fully open source AI model includes training data as a key element. However, some developers question how feasible it is to distribute the huge data sets required.
https://theconversation.com/what-is-open-source-ai-a-software-engineering-researcher-explains-236668
