I know this thread had most relevance prior to the elections but as a member of the academic community myself, a co-author of academic work on open source AI, and as an active contributor to the OSI ‘deep dive’ process, I must make exception to this.
In the academic sector, in my experience, it is much easier to get the idea across that openness can come in degrees and is a composite notion. I fully understand that OSI wants an all-or-nothing definition, but as a matter of fact, for academic and research purposes uses, even licenses that limit some of the freedoms (e.g. RAIL) are perfectly usable (as we’ve discussed before). In fact, given that the current OSAID makes compliance possible with an Apache or MIT license but without sharing data (and with a murky notion of “data information”), OSAID is not all that useful or important to academia, where access to data and the fullest possible information about data is much more important than open licensing.