Hacker News new | past | comments | ask | show | jobs | submit | miguelrochefort's comments login

I predicted those features weeks ago [1], but I still feel they could do so much more. If anything, their implementation feels like a rushed afterthought.

In their demos, they use the action button to capture an ambient song for Shazam, the power button to capture a voice command for Siri, and the camera button for an image for Visual Intelligence. All 3 captures should be performed using the same button.

And screenshots still require pressing 2 buttons simultaneously. Unless you want to share your screen with Siri in which case it's the power button...

People are going to use this button as a voice recorder, and Apple will announce native support next year.

[1] https://miguelrochefort.com/blog/capture-button


CS7637 was a fun course! One can get pretty far (80% accuracy) by using xor masks and comparing centroids.


How is this better than a custom launcher?


I highly resonate with this. In fact, I spent over a year buildng it [1].

My thesis is that the calendar will play a significant role in the next OS UI paradigm, replacing the old grid of icons and overlapping windows. Everything will happen around a unified timeline, through which you will launch an app by time-blocking its use, check notifications by looking at the past, forecast battery life or weather by looking at the future, undo/redo actions (or view snapshots/backups) by time traveling, etc.

It is shocking how far behind our map of time (e.g., Google Calendar) is compared to our map of space (e.g., Google Maps). Every spatial feature has an obvious temporal equivalent that just isn't implemented. I want to be able to search, save, and review events. I want to schedule an itinerary of events. I want turn-by-turn navigation in time.

Of course, solving time is just the beginning. The real magic happens when you combine space and time. For this, we may need to wait for more AR/VR adoption, whose added dimension should make this realization obvious to most and significantly facilitate its implementation.

[1] https://chronomize.com/


This was the idea of the Lifestreams project:

https://www.cs.yale.edu/homes/freeman/lifestreams.html


Thanks for your comment.

Your mention of combining space and time compels me to share my thoughts about time and space, I am experimenting with a programming technique that models state directly as its primary interface (whereas most programming languages model instructions). Time is inferred by movements between states or interactions with state.

I want "turn by turn" navigation of state and time as an implication of that.

It's incomplete but here is a point and click programming environment with some numbers and some parsed JSON into a grid. You can move data around between cells and the instructions are logged and the state is captured on each instruction. You "move through" state space. Try moving a value from the top to the bottom or vice versa.

https://replit.com/@Chronological/DynamicTables#index.html

Time travelling is implemented by clicking on instructions (the list items under Program instructions when you have done an action) The goal is to to infer the instructions needed to do the movement, be it pop off the stack or access some inner field in JSON or call a function. It's inspired by spreadsheets and the insight that most programming is logistics.


When you say "spent," does that mean it's in the past and you're not building it anymore? Or are you still actively developing it?

I love the idea behind it and would be very interested in using such a service, even if it might require some reimagining from traditional calendar format to make it fun and intuitive to use!



Startup idea: Use stable diffusion to automate corporate headshots and security badge photos. Charge $X to remove watermark.


"Notice of unemployment" (personal email address)


I would be surprised if we did not already have AIs that could invent calculus from first principles.

LLMs solve natural langage and common sense, which are not necessary for discovering calculus.


LLMs have not 'solved' either language or common sense. They will unknowingly and quite blatantly contradict themselves, for instance.


That's only because memories haven't been fully developed for AI, yet. Most LLM examples don't know what they said in response to input from yesterday, much less from 5 minutes ago.

Additionally, without a construct, like a body, in which to store these memories, AGI will continue to disappoint those who do.


You can propose the premise of your first paragraph, but it is not an established fact.

As ChatGPT can contradict itself within a single reply, I doubt that this is the only issue.


I believe our common desires to establish what is and what is not factual get in the way of innovation. Talking about something isn't going to solve the problem. Doing, might: https://arxiv.org/pdf/2201.06009.pdf


I'm extremely bullish on Large Language Models (e.g., GPT-3, ChatGPT) and Stable Diffusion (e.g., DALL-E).

This year, I will use them to double my productivity at work. I also plan to integrate them to my life management framework.


Say more? On how you plan to double productivity.


What is your current life management framework?


Here's a 24/7 background audio recorder app I made for Android. The impact on battery and storage is surprisingly reasonable.

https://github.com/miguelrochefort/eardrum


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: