@erikbryn@mattbeane@emollick Broader question raised: which has more leverage to resist AI adoption in the face of orgs with incentives to push AI—public or professional resistance? I suspect it will depend on their relative bargaining power. Open question for future work!
@erikbryn@mattbeane@emollick Love this! Would be interesting to see whether the public and workers are aligned in the *tasks* they want workers to delegate to AI. Unfortunately, we don’t have the granular task level ratings that Erik+team have to do the comparison.
@erikbryn@mattbeane@emollick We’re thinking about how we can build out this paper so that could be one direction. But a deeper consideration of how public resistance relates to professional resistance in Shao et al. is in the works for the next revision.
As someone who does basic social science research, it can often feel challenging to make findings interesting and relevant to people in industry, so it was particularly validating to know that I struck a chord with the company that forms an integral part of my dissertation.
Rounding out our candidates for today is @heatherjyang from @MITSloan. Heather's research examines resistance towards Artificial Intelligence, human and algorithmic stereotyping, and digital representation. Find out more here: https://t.co/GZkOaDqJxX
#jobmarketWOB
@AnthonyLeeZhang The Microsoft distribution also has some great reproducibility features baked in. Just specify the version you're using and it'll use the corresponding packages. Don't need to worry about updates breaking code!
Just recorded all my lectures for Quantitative Social Science Methods, I -- the first course in the Harvard Government Dept graduate methods sequence -- and am making them available publicly here: https://t.co/oQfRetbRZr I hope you find them useful for your course, or yourself
@LandonSchnabel Very similar to the way that I’ve been going about things. Thanks for sharing! The optimizer in me likes to think there has to be a better way of managing these kinds of nonlinear projects, but maybe the best way is really just to be flexible and embrace the uncertainty.
New ways of visualizing data aren’t given enough credit. Especially important for nlp and ml where models are more opaque. Difficulty interpreting results means these methods won’t see wider use among social scientists. Cool to see more work being done on this!