Yes that can be very useful, and can speed you up a lot. But someone must check the output.

If you let it operate on a prod system and it messed up, it's on you.

The core of the information they present isn't much different than what you'd hear on Dwarkesh or other industry podcasts, the presentation is some weird mix of ESPN and Mad Money that I personally don't get, but maybe makes sense to a US audience.

I don't see why that is interesting to OpenAI, but maybe I'm missing something.

I didn't go for a Max chip because I value the better battery life on the Pro more than I value the additional GPU cores.

Personally, I think until the LLMs start to plateau, it will always be more valuable to run a frontier LLM vs just a very capable local LLM. I have no idea when that will happen, so I simply decided to not overbuy the hardware now.

Because it's most likely in the training data. I.e., it stole it for you.

This commonly expressed non-sequitur needs to die.

First of all, all of the big AI labs have crawled the internet. That's not a special advantage to Google.

Second, that's not even how modern LLMs are trained. That stopped with GPT-4. Now a lot more attention is paid to the quality of the training data. Intuitively, this makes sense. If you train the model on a lot of garbage examples, it will generate output of similar quality.

So, no, Google's crawling prowess has little to do with how good Gemini can be.

Worktrees and parallel agents do nothing to help me with that. It's just additional cognitive load.

At least for my use, 200K context is fine, but I’d like to see a lot faster task completion. I feel like more people would be OK with the smaller context if the agent acts quickly (vs waiting 2-3 mins per prompt).