i believe that for most people on the street, for most tasks, a Chat GPT 3.5 era LLM is sufficient enough. sprinkle in tool calling and other things, and that becomes enough. if you can prioritize that level of a model on-device (baking it in etc), then you can bifurcate AI users between those unwilling to pay and those who are willing to pay A LOT for frontier model performance.
there’s no hard evidence here. the “99%” referenced in the article is someone’s personal subjective confidence it’s him. body buried under church is not particularly eventful news as it stands.
isn’t what cook is pointing out actually the most important thing?
a product created for the strategic purpose of expanding into a new clientele is doing exactly that. that is the win.
put another way, if the statement he said was “best launch week ever for Mac customers.” that does not speak to the entire reason for the existence of this product category. in essence, THAT would be the pointless statement.
uh isn’t the data leaker the necessary accelerant and necessary component to validate against the rest of the ecosystem? isn’t that what triggered the communication and coordination between multiple delve customers?
putting a build on your own apple watch is horrific.
it constantly disconnects, requires restarts and other nonsense techniques. i legit do not know how you can not be running into these problems if you are developing on those platforms.
we saw this exact playbook with ios 7. i don't think you need to attribute malice or read into it much.
ios 7 relied heavily on blurring effects-- a flex at the time due to the efficient graphic pipeline vs android they had. this was coming off the heels of Samsung xerox'ing and they wanted a design that would be too expensive for competitors to emulate without expensive battery hit. liquid glass is following in this tradition.
and similarly to ios 7, the move to flat design was predicated on the introduction of new form factors and screen. flat design lent itself well to new incoming screen sizes and ratios. prior there was really one or two sizes and that was it, easy to target pixel perfect designs against. as apple moves to foldables and more, this design flexibility is once again required.
as for no one trying to emulate it, i'm not so sure, OriginOS 6 ripped it off back in October.
i believe that for most people on the street, for most tasks, a Chat GPT 3.5 era LLM is sufficient enough. sprinkle in tool calling and other things, and that becomes enough. if you can prioritize that level of a model on-device (baking it in etc), then you can bifurcate AI users between those unwilling to pay and those who are willing to pay A LOT for frontier model performance.