Traditionally, all computer science research papers underwent a review process to determine whether the paper met the publication criteria for trade research journals. The reviewers were professors and senior researchers who decided whether the paper made a significant contribution to the industry. This “program committee” served as editors for the scientific community. However, the advent of arXiv has changed this tradition, says Jason Mars, a professor of computer science.
ArXiv is a free open access e-archive that allows scientists in physics, math, computer science, and related disciplines to share articles globally before the research is published. While some of the papers self-published on ArXiv are insignificant, many of them represent top-notch research work, says Jason Mars.
“The most impactful papers in AI (artificial intelligence) were published on arXiv before ever being published in a peer-reviewed venue,” says Jason Mars. Publishing on arXiv allowed the AI field to progress faster than it would have if the scientists had waited for peer-review without publishing, he says. “ArXiv is first to market with everything.”
New technologies are coming out so quickly that the possibility of an idea getting scooped is increasing. Researchers are pressured to put their ideas out quickly, says Jason Mars. Peer review, however, takes several months. “The general sense of the community is that “ain’t nobody got time for that!” says Jason Mars. The result is that scientists turn to ArXiv to put their ideas out quickly.
When scientists self-publish on ArXiv and open-source their research code, others also can run the code to test it to see if the new technique is better than the old method. In this way, the crowd becomes the reviewer, says Jason Mars, resulting in an economy of ideas and a competitive market for contributions. In this way, ArXiv has disrupted the space. “No longer will we have a group of greybeards deciding on what ideas hit the world,” says Jason Mars.
As the world becomes ever more digital, this trend will exacerbate, and “I say good riddance to the old and in with the new,” says Jason Mars. “The economic dynamism of the market is the most durable meritocracy that has yet emerged.”
About Jason Mars
Jason Mars is a computer science professor at the University of Michigan, where he directs Clarity Lab. Clarity Lab is one of the best places in the world to be trained in topics spanning artificial intelligence, large scale computing, and programming languages. Among numerous contributions, his lab pioneered award-winning work on Sirius (aka Lucida). Sirius is the world’s first-open source sophisticated virtual assistant that simultaneously hears, sees, and understands.
Jason Mars also founded Clinc, a record-breaking cutting-edge AI company based in Ann Arbor, Michigan. At Clinc, he led the development of the world’s most advanced conversational AI, which was first applied to the banking industry and then later to the healthcare, automotive, and food services industries. From 2015 to 2020, Clinc experienced substantial year-over-year growth employing 120 employees and raising more than $62 million in venture capital. Under his leadership, the business has grown to about $10 million in ARR, and its AI products have more than 7 million active users.