@adamhjk@stack72 I mostly agree, but I do think it's much less human effort to get non-idiomatic, average quality language translation than an idiomatic, high quality refactor
@stack72 Nice post. Mechanically, how do you see encoding working? Agents take incident postmortem text from slack and model any currently unmodelled system constraints, inclusive of any new types that need to be designed? I have only read the front page of the swamp site.
@jessethanley@lesstenny CGT discount on property addresses housing affordability, and this seeks to advantage wage earners over share holders. It's a blunt tool but that seems to be their approach.
@PatrickHaede@mwseibel@m16vie Here in Australia, where there are massive housing supply issues, Airbnb and similar short stay rentals further distort the market. It's nice that people can feed their family this way @mwseibel , but do remember there's another family on the other side of this transaction.
@PatrickHaede@mwseibel@m16vie Here in Australia, where there are massive housing supply issues, Airbnb and similar short stay rentals further distort the market. It's nice that people can feed their family this way @mwseibel , but do remember there's another family on the other side of this transaction.
@TomDavidsonX@mentalgeorge@Manderljung Could AI researchers make that breakthrough? To me the answer is no, models can only be creative within the distribution they are trained on, and this would be extrapolation, but I also don't know how humans (say, Picasso) can seem to be.
@TomDavidsonX@mentalgeorge@Manderljung I do personally agree, but then the timing depends on an algorithmic breakthrough that leads to sample efficiency. To me that feels bigger than previous breakthroughs like Transformers and GANs, noting that some researchers believe they have a way forward e.g. world models.
@iamtrask Professionally (swe), it's slow adoption due to risk aversion - that secrets might be leaked or vulnerability introduced. Personally? I think that I have changed what I spend time on as a rejection of the kind of busy work that models excel at, if that makes sense.
@bgurley This was an experiment getting Claude to help me do frontend dev while I wrote the content. Very satisfied with how easy it was to accomplish a lot in a short amount of time. My frontend skills are weak and I am not sure making them stronger is the right direction for me.