00:23:19	Orit Shaer:	Questions? Type them here:
00:26:05	Peter Mawhorter:	I’ve been trying to use the standard of informed consent in my recent AI training research (which so far effectively has meant forgoing most public training data sets entirely). Do you think that this is a useful standard to try to apply to training datasets, and is reducing the exploitation of “freely available” data a useful supply pressure to help increase the value of labor for AI purposes?
00:27:46	Albrecht Schmidt:	How does this relate to "old" ideas of Unions?
00:28:01	Carlos Álvarez:	Location dependent gig platforms, such as courier delivery, have started to explore platform collectives. Can you elaborate on what worker-owned crowd work platforms could look like?
00:30:57	Aakash Gautam:	Would we not perpetuate the problems (e.g. inequity in distribution of benefits and harm) by presenting gig marketplaces as the alternative for the gig workers?
00:37:01	Sandy Gould:	When I talk to my colleagues in the Business School about upskilling crowdworkers, they tell me that if we increase their productivity then ultimately wages will correct (i.e., they will get paid less per unit work). This paints a pretty depressing picture to me! What’s your take?
00:40:14	Peter Mawhorter:	(…that’s a general problem with capitalism as a whole, fo course)
00:41:19	Albrecht Schmidt:	can we do something that customers care about the workers they do not see?
00:41:34	Peter Mawhorter:	(That does seem like a key issue here!)
00:42:08	Aakash Gautam:	Another related example: Initially Uber drivers (when there are few) earn more and hence are incentivized to invest (in new(er) cars for example). Later, as number of drivers in the area increase, the rate of return would lower.
00:42:22	Sandy Gould:	Thanks, Saiph! Sounds like there’s an interesting empirical question here.