The history of artificial intelligence has shown time and again that openness breeds progress. When Google open sourced TensorFlow and the Transformer model in 2015, it catalyzed innovation across the AI community. Suddenly researchers everywhere could build upon these tools to create their own breakthroughs. The incredible advances we’ve seen in AI over the past few years would not have been possible without this spirit of open collaboration.
Yet some are now retreating back into silos, hoarding their research and models out of fear of losing competitive advantage. This may benefit individual companies in the short-term, but it is detrimental to the field as a whole. As Yoshua Bengio argues, AI is a “positive-sum game” where openness allows collective value creation. All boats rise with the tide.
Take Meta for example. By open sourcing key technologies like PyTorch and React, they have simultaneously advanced their own capabilities while empowering others. The results speak for themselves – Meta has become renowned as one of the leading innovators in applying AI. Openness has given them a competitive edge, not taken it away.
The lesson is clear: open sourcing accelerates progress in AI to everyone’s benefit. As Geoffrey Hinton puts it, “if you’re not in the open-source movement, you’re not really in the AI movement”. The future health of the field depends on a continued commitment to open research, open data, and open models. Only through open collaboration can we ensure AI develops safely and for the common good.
So let us embrace the open source ethos. Share ideas and code freely. Publish models and datasets. Work together across institutional boundaries. And collectively build an AI future we all want.