Microsoft, Google, and Amazon also do this, but they also have the hyperscaler datacenter infrastructure to host the chips. Designing and taping out the chip is one thing, packaging, cooling, deploying, powering, and managing the fleet is another stack entirely. Wonder where that will come from?
I drive a 2004 truck, it's not as big as the newer trucks but it's still pretty big. My 2004 truck also has 0 sensors on it and no cameras. If I were driving a newer truck, I believe pedestrians would be safer because of the added safety sensors and cameras.
Why would Apple’s inference fleet be more efficient? They would need to get into the TPU/water cooling business to make that happen. Apple’s silicon is great but remember it’s based on ARM an already existing ISA with reference designs. The burden to make Apple’s CPU silicon is lower than what it would be to make a TPU.
Have to remember that regulated gambling also has lobbyist so unless regulated gambling is invested in these apps they are likely trying to get them banned or also regulated
Flip the classroom, make students learn the material on their own (Using AI or whatever resources they want to use) and then in-classroom time is divided on working on problems (without AI assistance which can be controlled in this environment) and quizzes/exams (again without AI). We don't need lectures anymore, they are an incredibly ineffective way to learn.
Really I see it as the exact opposite. A walk around the neighborhood is highly unlikely to strike up random conversation with a stranger. Might run into a friend or someone friendly but they are likely going to approach with or without airpods
on prem cloud is harder because of the scale up and scale down requirements. If you are a growing business which most decent ones are, you constantly have to think about that.
Main use for consumers / tinkers. In industry, the main use is during pre-silicon development. You emulate the target processor and model the periphs and you have a complete virtual representation of your new SoC/MCU/etc before the hardware is available. Benefit being once you do get the hardware, you already have the entire software stack nearly ready to go and already tested.
"training good coding models" many would say that is a highly debatable statement, and some would say that is just flat out not true. Cursor has not trained a frontier model from scratch, what they did was take an already made (non-frontier) model and further trained it on their user data about coding outcomes from its coding agent. So, a form of distillation and RL.
LoC by itself is useless and so is AI LoC, it doesn't really show anything by itself.
But if you pair AI LoC in a range and also task completed in the same range and then compare that with historical data over a similar range without AI, then you have something tangible.
You also need to look at defect reports to understand the full picture of is AI being helpful.
So, we do need to measure AI LoC and AI PR counts, but we also need to make sure we are using other metrics to help paint the full picture.
I think a lot of developers who enjoy writing code are in an internal battle knowing they could write it themselves or AI could do it and probably do at same speed or many times faster. It's a contradiction as a developer because we always want to do whatever it is we are doing at the optimal speeds but we also like writing code. So, you're being forced to choose where you stand at a fundamental level.
I think it’s pretty easy to judge when in conjunction with AI edited photos you have algorithms that award that. So sure it might be just some people that do it, but it ends up being all that is seen. Is Apple to blame in that? No. But are they completely innocent? Also no
T1 companies have longer depreciation cycles, they have customers that will use the dated hw for non-frontier work. They can make the capex more justifiable and have flexibility to be more creative about its use. A frontier lab really needs the best hw available at full capacity.
Not sure I agree with the cropping argument, would be no different than cutting a printed photo with scissors. I would lump the other image manipulations done by hardware into "AI manipulation" as well, although those tend to be much more modest.
Think two things can be true at once. They should be using their capital to achieve their speculative price. Instead, they are using their capital to achieve a modest ROI, thus invalidating the speculation AND proving they have tech issues in what the speculation is around.
I have to say, I extremely dislike AI processing of photos. The camera is a vehicle to capture the realness of the world around us, including the imperfect moments. Distorting that with AI and being okay with it is really disappointing.
Congress is allowed to make laws (covered by the constitution) if that law grants the federal government the authority over something then the law is covered by the constitution.
In terms of our brains though we can only think forward as well (if forward is time). Our brain in the future says something we did in the past was wrong (part of the sentence we wrote) and that informs our body (the agent) to go back and fix it