Artificial Intelligence, a new culture for technological advancement
We have discussed in previous articles that the advances in artificial intelligence (AI) in this century have been favored by the convergence of three technological factors: the availability of large volumes of data, communication networks that connect large areas of the planet at high speed, and powerful computers to process all the information. These factors have provided the infrastructure to implement the systems and algorithms that surprise us day by day.
However, two other elements have increased the speed of advances and the dissemination of knowledge, as well as reduced the time that applications arrive on the market. The first is the open culture of the 21st century, which advocates the free dissemination of knowledge and information. The second is the leadership of corporations in AI research and machine learning (ML). Both elements, along with the technological factors, produce a dynamic unusual in technological advancement processes.
The dynamics begins with the presence of leading researchers, authors of the most important recent developments, within the main corporations. Geoffrey Hinton, the initiator of the rise of deep neural networks, is on Google's research team; Yann Lecun, pioneer of the use of convolutional networks for computer vision, leads Facebook's AI team; Ian Goodfellow, creator of the generative adversarial networks model, is now ML director at Apple; Spyros Matsoukas, known for his voice recognition work, leads Alexa's development at Amazon, and Eric Horvitz, a decision-making specialist for AI agents, runs Microsoft's research labs.
This is because research papers and the conferences where they are presented have become a kind of showcase, in which companies detect and attract researchers with projects that have the potential to strengthen their AI and ML strategy. This practice has generated a great interest in writing and publishing papers, which in turn has caused a disruption in the media where they are published.
Research papers are traditionally published in specialized journals, where the articles proposed by the authors are reviewed by other researchers, who have their opinions on the acceptance of the article and/or comment on it in a review process that takes several months. These publications have great prestige and most of them are of restricted circulation and high price.
The dynamics of using papers as a portfolio to make oneself known, the diminishing lifespan of a paper because of the speed with which new advances are made, as well as the willingness of many researchers to seek the free dissemination of their papers, have led specialized journals to change, as other means have emerged to publish papers that respond to the need for speed and wide dissemination demanded by authors.
These new media are manifestations of one of the most positive aspects of internet and social media interaction, the culture of openness and crowdsourcing. This culture promotes the free dissemination of knowledge and information, not only for a large number of people to benefit from them, but to bring out new knowledge and information that enrich the collection of knowledge, as happens in the extraordinary phenomenon of Wikipedia, the free encyclopedia maintained with the effort and donations of thousands of people.
In line with this open culture, big corporations in the industry maintain internet sites where they publish many of their research (understandably that they don't include those that give them a competitive advantage). The same is done by the main universities and even sites specialized in the compilation of articles have emerged.
In this way, thousands and thousands of research articles on all the imaginable topics and applications of AI and ML are downloaded and read by academics to take as the basis for new jobs, by teachers and students to learn, as well as by professionals to apply them in their projects. This dynamic promotes the rapid dissemination of advances and the emergence of new ones, as well as facilitating the implementation of applications in businesses.
But the culture of openness and collective effort is not limited to academic research, some collaborative software development platforms have been very successful as tools to develop AI, ML and data science applications, or to create a professional prestige with a portfolio of projects. On these platforms a person uploads a project, with code and data, that receives feedback or modifications from other people on their team, if it is a company or team group, or anyone in the world if open access is given.
An open source project allows anyone to use it, but in return it receives feedback from a lot of people. The model is so promising that on some platforms there are contests with cash prizes, in which the specifications of an application to be developed, or a set of data to be analyzed, are uploaded, and numerous proposals are received, of which we choose the winner of the prize. An effective way to have many developers looking for solutions to our problem.
The development of AI is not without risks. However, it is encouraging that a culture of openness and collective effort prevails. It not only accelerates technological advancement, but the dissemination of knowledge and collective effort mechanisms facilitate access to the technology for more companies and individuals, while providing a public surveillance platform for AI’s development.
Here are some links to sites with research articles and development platforms.
Sites that publish papers
Did you enjoy this post? Read another of our posts here.