Thanks for flagging this Neil, good find! We’ve put this on the medium/low priority list for ARC & METR, since the EV is pretty low given our donors (but I think positive!).
One little puzzle: given that METR is trying to stay independent from major AI labs, do we reject donation matches from Google?
Is there any situation you predict in which Google donation matches would affect METR’s vision? What is the probability of that happening, and what is the value of donations to METR by Google employees?
If you’re asking for advice, it seems to me that refusing donations on principle is not a good idea, and that donation matching from Google for employee donations carry no legal bearing (but I have no idea) and are worth the money. Besides, I understand the importance of METR independence, but are Google and METR’s goals very orthogonal? Your final calculation would need to involve degree of orthogonality as well. I’m not a very valuable data point for this question, however.
Thanks for flagging this Neil, good find! We’ve put this on the medium/low priority list for ARC & METR, since the EV is pretty low given our donors (but I think positive!).
One little puzzle: given that METR is trying to stay independent from major AI labs, do we reject donation matches from Google?
Is there any situation you predict in which Google donation matches would affect METR’s vision? What is the probability of that happening, and what is the value of donations to METR by Google employees?
If you’re asking for advice, it seems to me that refusing donations on principle is not a good idea, and that donation matching from Google for employee donations carry no legal bearing (but I have no idea) and are worth the money. Besides, I understand the importance of METR independence, but are Google and METR’s goals very orthogonal? Your final calculation would need to involve degree of orthogonality as well. I’m not a very valuable data point for this question, however.