
Neural networks and artificial intelligence systems are a hot research topic these days. Unfortunately, these studies are often conducted in the depths of commercial companies, such as Google and IBM, and are closed to the public. To remedy the situation, you need to conduct research openly, according to enthusiasts from the Free Software Foundation, who launched the
GNU Gneural Network initiative.
The Gneural Network package is a programmable neural network. In the current version 0.0.1, this is a fairly simple network of direct distribution, which performs simple tasks like the selection of formulas — plotting curves on a set of values, but very soon the developers are planning to significantly expand the functionality. Right now we are working on the implementation of the LSTM (long short term memory) architecture for a recurrent neural network and deep learning. There are also plans to implement training with reinforcements.
The Free Software Foundation (FSF) emphasizes the ethical motivation of this project. Artificial intelligence corporations such as Watson and AlphaGo are truly impressive. But only employees of research laboratories of these companies have access to them. The FSF sees this as a threat. First, it is not known how companies are going to use AI, if their main goal is to make a profit. Secondly, the monopoly on scientific research hampers scientific and technical progress.
')
“In fact, data and trained neural networks are rarely published, and it must be remembered that in practice it is almost impossible to repeat the result without training data (even if the source code is available),” writes FSF. - We have to ask ourselves the question: do we really want the AI ​​to be available only to the elect? Shouldn't Progress and Knowledge be open to all? ”
In the light of the possible emergence of human-level AI, there are concerns that a “commercial” AI will act not in the interests of humanity, but
in the interests of corporations . Gneural Network is not the first attempt to translate experiments in the field of AI into open access. Recently, Ilon Musk with partners created a
non-profit organization OpenAI . The
official announcement says: “Due to the unpredictable history of AI, it is difficult to predict when AI of the human level can appear. When this happens, it will be important for mankind to have a leading research institute that is able to prioritize the gain for all over their own interests. ”