Actually many people criticizw Apple's AI power, but they forget that the reason Siri is stupid because Apple want to (or at least they explicitly show) protect your privacy.
Iphone AI core is basically Edge AI, which means they deploy it inside the Iphone, not relying on a remote cloud. This can help protect customer data, but it also means they can not update their AI with new training data, and the Chip is kinda small so the model is not smart.
Just know this fact so ppl won't criticize Apple without knowing the sacrifice to protect their privacy
It's kinda useless to raise this point unless we can think of some action to do about it.
Should people avoid the sun? Probably not. Sunlight is proven to have a lot benefit in regulating mood and biological clock.
Should people wear sunscream. Maybe yes, maybe no. I'm not sure. I myself prefer sunlight touch my skin. So unless the UV rate is too high, I won't put on sunscreen.
What are the actions that normal everyday ppl can take in your opinion?
I guess one comment I have would be related to value propositioning. I read through your site, haven't tried it yet, but I don't immediately see why would I use it. Like "Create AI Blog that learn" is too vague to know what your product actually help my usecase. What does it learn, what problem does it show.
I think we shouldn't over think this kind of stuff. Just obssessed with being so good at what you do, occasionally go to clubs or events in the fields you are interested, and people will notice you if you're good.
In my case, I was simply invited by people who already know my ability. I don't even actively find it.
Interesting idea, but I don't see much use case for it. Is it gonna be an another AI-powered photo editing feature, just like Photoshop or Canva, but more tailored for real estate? I understand that you want to make these eal estate photos look more attractive? But what's that point if the real places don't match the ads?
That's what I've saying forever. All of a sudden, everyone's thinking being a software engineer is just all about writing code. In fact, coding is easiest thing for devs. And to be honest, I think everyone's misunderstanding the term "vibe coding" so much that if you use any AI tools in your workflow, you're labeled a vibe coder. That's not Andrej karpathy meant when he coined this term.
Love your work, but can you share what criteria these llms use to evaluate prompts and it would be nice if these llms can give feedback on improving these prompts beside giving just a score!
Love your work, but pairwise scoring also skips the bigger group context—benchmarks vs list‑wise or MMR methods would highlight trade‑offs. And I’m curious what the compute and latency hit looks like when you run all those pair comparisons in production.
I love the vision of a privacy‑first, open‑source browser that runs AI agents locally, but shipping a custom Chromium fork is a huge maintenance lift so how will you handle timely CVE patches? And the demo video on your website is a bit too generic, can you show a more task-specific one?
I hear your concern, and you're absolutely right to be thinking about this. It's true that AI development involves intense competition, and that can feel overwhelming - like we're all just passengers on a train we can't control. But I don't think we're as powerless as it might seem.
Throughout history, we've found ways to guide powerful technologies toward better outcomes, even when there were strong economic incentives pushing in other directions. Think about how we've developed safety standards for cars, regulations for medicines, or international agreements around nuclear technology. It took time and effort, but people working together made a real difference.
When it comes to AI, we still have meaningful choices. We can support leaders who take these issues seriously, back companies that are genuinely trying to develop AI responsibly, and speak up when we see problems. We can also stay informed and help others understand what's happening - sometimes the most important thing is just having honest conversations about what we want our future to look like.
What counts as “attending” when AI is present?
If a bot represents you in silence, is that equivalent to skipping the meeting—or attending? This raises deeper questions about participation equity and presence.
What validation approaches are on the table? A syscall extractor + LLVM libFuzzer? Harnessing KVM/QEMU for syscall replay? It’d be incredible to see how practical these tests are integrated into CI pipelines.
Congrats on shipping this—memstop looks incredibly useful for anyone diagnosing memory usage regressions in long-running Python apps or ML workflows. I especially appreciate the simplicity: one function call, no external agents, and integration with stdlib tracemalloc for context. Clean and focused.
Any plans to support tracking across async tasks or threads?