|
|
|
|
|
by steev
2155 days ago
|
|
I don't see how this is a nuanced perspective - it seems to restate the same complaints/arguments just about every comment makes in these discussions. A nuanced perspective would look at the arguments as to why OpenAI is doing the things they are doing. For example: * OpenAI publishes in closed journals (actually conference proceedings) because that is where all the cutting edge research is published and reviewed. I cannot recall an OpenAI paper that wasn't available either via arXiv or their website, despite being published in a closed journal. What is the alternative here? Where should they go for quality peer-review? Yes you can argue the peer review at top conferences is not quality, but is worse quality than no peer-review or peer review from open-access no-name journals? * How does OpenAI make money? How much are they bringing in? How much does it cost to support things like the OpenAI Gym, etc.? How much does it cost OpenAI in terms of bandwidth to host pre-trained versions of GPT-3? At some point a company needs to make money and prioritize resources - they can't give everything away for free in perpetuity. I don't think these questions have obvious answers - there is give and take. |
|
Organizations are constantly making decisions that are trading off certain values for others, i.e. openness vs safety/expediency/funding. But if they use the word open in their name, signalling to people that is one of their foundational values, people will expect them to pick openness even when it's not necessarily the easiest, safest, most expedient, or most profitable choice. They expect them to pick openness when it's hard.