Re:AI at Amazon: A case study of brittleness
by simplesimon890 on Wednesday June 11, 2025 @11:50AM
(#44248898)
Unfortunately it aligns with internal pressure. If you are not working on something that has clear quantifiable/promotional benefits that can be realized within 2-3 quarters, you are at risk of the PIP train. Having buy in from senior management can help but in a company that re-orgs regularly, your manager can change, so the risk is higher.
Re:AI at Amazon: A case study of brittleness
by gloryjulio on Wednesday June 11, 2025 @04:35PM
(#44251530)
^This. If this is only about less pay, lots of people would actually do it instead of chasing bonuses and promotions.
But in a cut throat environment, you can't afford to not move the corporate metrics in quarterly reviews. Otherwise you will get pipped or fired
Re:AI at Amazon: A case study of brittleness
by wnevets on Wednesday June 11, 2025 @02:55PM
(#44250629)
> No incentive meant no action taken.
sounds like late stage capitalism
Re:AI at Amazon: A case study of brittleness
by reliabilityguy on Wednesday June 11, 2025 @04:18PM
(#44251358)
> sounds like late stage capitalism
Yes. Because with socialism and communism the management cares about everyone’s benefits and not their own skin?
Re:AI at Amazon: A case study of brittleness
by SoftTalker on Wednesday June 11, 2025 @03:03PM
(#44250706)
Sounds like simple poor management to me. If you're going to do R&D work or invent something you hope to be able to sell, you have to accept that value may not be realized immediately and there may be some paths explored that don't end up leading to anything.
If you're going to sell T shirts, then sure, have quarterly goals.
Re:AI at Amazon: A case study of brittleness
by wnevets on Wednesday June 11, 2025 @03:17PM
(#44250822)
> If you're going to sell T shirts, then sure, have quarterly goals.
The issue provided in the example is about helping other teams out without an expectation of a reward, not about short term vs long term gains.
Re:AI at Amazon: A case study of brittleness
by nine_zeros on Thursday June 12, 2025 @02:30AM
(#44254778)
Yep. Fundamentally, tech R&D requires management that will be ok with trial and error, and ambiguous timelines.
The predictable unit production/sales way of management works for duplicatable, repeatable goods.
> And most importantly, there was no immediate story for the team’s PM to make a promotion case through fixing this issue other than “it’s scientifically the right thing to do and could lead to better models for some other team.” No incentive meant no action taken.
Oof!