The launch of Cohere: an open-source lab
The formal launch of Cohere For AI, a nonprofit research lab and community, has been announced on June 14, 2022. The lab will focus on tackling some of the most demanding issues in machine learning by delivering open-source, basic machine learning research. Sara Hooker will lead Cohere For AI, bringing with her a plethora of experience in AI and machine learning, focusing on deep learning. Sara worked as a Research Scientist at Google Brain before joining Cohere For AI. She focused on training models beyond top-line metrics to successfully show interpretability, compactness, fairness, and robustness. She also started Delta Analytics, a nonprofit that brings together researchers, data scientists, and software engineers to offer their time and expertise to nonprofits worldwide. "We need to make sure we're collaborating across a wide collection of individuals, disciplines, backgrounds, and locations to realise the potential of machine learning," stated Aidan Gomez, CEO and Co-founder of Cohere. Within the AI community, there has long been a fear that insufficient money is being set aside for AI research outside big firms. According to one study, corporate links to AI research — whether through financing or association — increased dramatically from 2008 to 2019. Another research found that between 2004 and 2018, Google parent firm Alphabet and Amazon and Microsoft employed a total of 52 tenure-track AI academics, displacing these potential lecturers from university and nonprofit positions. Concentrating power inside businesses has a lot of apparent drawbacks, but one of the most concerning is that it tends to undervalue ideas like benevolence, fairness, and participation in research. In an interview with Wired for a 2020 story, several experts argue that corporate AI efforts have resulted in an "unscientific focus" on projects that are only achievable for those with access to big data centres.
Reasons to open the non-profit Lab:
Work within organisations, regardless of area, is sometimes highly guarded, taking years to reach the light of day – if it ever does. Access to computational resources is another key component Cohere For AI is seeking to improve, according to Hooker, to assist researchers in better utilising "cutting-edge" models to help advance their work. " The role of computing access is evolving, as evidenced by language model developments (i.e. AI systems that understand and generate text). A few years ago, developing a complex language model needed a large amount of computing power. The obstacles to the entrance are now much lower than before due to scholarly advances and the work of the open-source community. In addition to foundational research, Cohere For AI will host a machine learning community where members may interact with one another, meet new people, and engage in open conversation and collaboration. The lab and community will aim to provide new points of entry into machine learning research and, in the end, will represent its members' diverse experiences and interests.
About Cohere:
Cohere For AI is a nonprofit research lab and community devoted to advancing machine learning research and solving some of the field's most difficult issues. It promotes ethical machine learning research while emphasising appropriate stewardship of open-source scientific approaches. Cohere For AI is a community-driven research lab inspired by the possibility of building an inclusive, dispersed community made up of exceptional research and technical talent from across the world.
