"remember when Apple and Microsoft were the mortalist of enemies?" Oh yeah, biggest memory around that is when Apple were in somewhat dire straits (late 90s), Steve Jobs was presenting at Macworld and Bill Gates came up on the big screen behind him to announce a partnership with Microsoft and a $150 million investment in Apple - and the crowd booed heavily. Apple is way behind on AI. Leveling up Siri is such a very very low bar. I think their best shot at the moment of getting 'back in the game' is if they make a real on-device model (not one that has to go out to the web via ChatGPT for anything not 100% basic) and make it smart enough to take the onus off users to know which app, or set of apps, to use to carry out common tasks - and then less common tasks etc.
Trying an open source model is a great idea. Just my 02, I don't want Llama anything because it's part of Meta and not fully open source, Qwen 3 is China made so ... I would use Gemma or a Mistral model or even one of Microsoft's PHI models
do you use a Wacom tablet or something else to draw? The new iPad doesn't support the old Apple pencil, so if I want to keep doodling in Procreate I'll have to spend more :/
I’m totally meh on Apple Intelligence as well. I’ve been an iPad guy from the start … and it remains great for reading and writing, but for me not a whole lot more
Looking into this a little more, I learned the new chip will let you run models locally (that's the 'neural engine' stuff) and it's meant to make Siri amazing sometime next year. Maybe I'll play around with it - supposed to be super fast and super secure - no internet required.
"remember when Apple and Microsoft were the mortalist of enemies?" Oh yeah, biggest memory around that is when Apple were in somewhat dire straits (late 90s), Steve Jobs was presenting at Macworld and Bill Gates came up on the big screen behind him to announce a partnership with Microsoft and a $150 million investment in Apple - and the crowd booed heavily. Apple is way behind on AI. Leveling up Siri is such a very very low bar. I think their best shot at the moment of getting 'back in the game' is if they make a real on-device model (not one that has to go out to the web via ChatGPT for anything not 100% basic) and make it smart enough to take the onus off users to know which app, or set of apps, to use to carry out common tasks - and then less common tasks etc.
I was there for that :) on this new hardware I can run Llama 3, Gemma 2 or Qwen 3 locally. Should I try it 🫣
Trying an open source model is a great idea. Just my 02, I don't want Llama anything because it's part of Meta and not fully open source, Qwen 3 is China made so ... I would use Gemma or a Mistral model or even one of Microsoft's PHI models
I'm still "nursing" a 10+year-old Dell Precision T3600 workstation running WIN10. Her time will come, but the old workhorse is still doing OK for me.
do you use a Wacom tablet or something else to draw? The new iPad doesn't support the old Apple pencil, so if I want to keep doodling in Procreate I'll have to spend more :/
I’m totally meh on Apple Intelligence as well. I’ve been an iPad guy from the start … and it remains great for reading and writing, but for me not a whole lot more
Looking into this a little more, I learned the new chip will let you run models locally (that's the 'neural engine' stuff) and it's meant to make Siri amazing sometime next year. Maybe I'll play around with it - supposed to be super fast and super secure - no internet required.