[{"id":1198,"title":"'Accuracy per FLOP': A Green AI Metric for Fair and Efficient AI Development","sha":"7ca347b740b1e3876e3987d1fc986df9","user_id":null,"state":"published","body":"Recent advancements in Artificial Intelligence (AI), particularly through Deep Learning and large-scale datasets, have led to the creation of powerful models like OpenAI's GPT-$n$. These models have significantly enhanced Natural Language Processing and driven innovation in the field. However, concerns are growing about the monopolization of AI by entities with the resources to develop such large models.\r\n\r\nThe current emphasis on accuracy favors well-funded projects, potentially stifling innovation and disadvantaging researchers with limited resources. An alternative metric, “accuracy per FLOP” (floating-point operations), could provide a fairer comparison by measuring model efficiency relative to computational cost. The concept of “accuracy per FLOP” aligns with \"Green AI,\" a term from [Schwartz et al., 2020](https://doi.org/10.1145/3381831), which advocates for considering computation cost in AI research.\r\n\r\nTo promote inclusivity and innovation, researchers should explore new techniques and models that improve efficiency rather than merely increasing model size. This shift is crucial to ensure AI research benefits everyone and avoids a potential decline in AI progress, known as an Artificial Intelligence Winter. Human-centered AI, focusing on human interaction, should also remain a primary research focus.","created_at":"2024-05-31T09:46:53.143Z","updated_at":"2024-09-30T07:53:42.999Z","zenodo_id":12660216,"doi":"https://doi.org/10.5281/zenodo.12660216","subject":null,"tags":["artificial intelligence","green ai","democratization","new metrics"],"vote_count":2,"view_count":1,"score":2.1,"deleted":false,"muted":false,"tweeted":false,"attachment_file_name":null,"attachment_content_type":null,"attachment_file_size":null,"attachment_updated_at":null},{"id":1432,"title":"The need and format for publications is stifling innovation and science adoption","sha":"326de755c8bdb494b354e804fb09a74e","user_id":null,"state":"published","body":"Nowadays two problems affect science.\r\n- The *quantum* necessary to publish scientific discovery is large. This not only heavily increases the time needed for publication, but it also impacts the time required to stay up to date with the scientific literature, stifling innovation. This is a problem generally accepted and recognized: in fact, the \"Journal of Brief Ideas\" was created with this exact goal in mind ([Harris, 2015](/ideas/17b997da912f99ce9988a47f42b52692)).\r\n- Back when science was just starting out, scientists were used to publishing for the general public. In fact, most of the *classics* books on \"science\" can be read by everyone, although the general, lay reader may find some difficulties in this activity. This probably increases the chances that science is discovered, and adopted.\r\n\r\nWould it be possible to evaluate the impact that the standards of publications have on science? If so, how? Should researchers be more \"conscious\" of the general public, allowing for broader access to science and potentially improving our overall culture? Should researchers be encouraged to disseminate their research to the general public? If so, how should these changes be proposed to academia?\r\n\r\n_**Disclaimer:** I got this idea while reading: [\"How to Read a Book\" by Adler, M. \u0026 Van Doren, C.](https://www.google.com/books/edition/How_to_Read_a_Book/Xg-CBAAAQBAJ?hl=it\u0026gbpv=0\u0026kptab=overview). Further research is needed to confirm the hypotheses._","created_at":"2024-07-29T16:10:23.342Z","updated_at":"2024-09-19T02:10:16.965Z","zenodo_id":13129574,"doi":"https://doi.org/10.5281/zenodo.13129574","subject":null,"tags":["publications","science adoption","opinion"],"vote_count":1,"view_count":1,"score":1.1,"deleted":false,"muted":false,"tweeted":false,"attachment_file_name":null,"attachment_content_type":null,"attachment_file_size":null,"attachment_updated_at":null}]