Reads similar to https://en.wikipedia.org/wiki/Laplace%27s_demon

similar in the sense that they're both influenced by determinism.

funny enough, i wrote a draft blog post about what it would take "in theory" to be able to predict the future, but never got to finish it.

I was really intrigued by the title but "without installing anything" means "installing an extension" or not?

I'm waiting for this! But I'm to consumer / usage is rare for me to build from repo

how often did I pickup my phone and just want to say "Hey my parents are over for my 5yo's birthday can we have something upbeat while we eat cake" and then be able to follow up "less vocals and a bit slower" to adjust the mood.

Yeah, I do understand you.

I do have plans to build a working website for this

Currently I am in need for a frontend developer because I am not a full stack dev, just Backend and AI Engineering for now

So could SaaS LLM or cloud/api LLMs not offer this as an option? A guarantee that the "same prompt" will always produce the same result.

Also the way I usually interpret this "non-deterministic" a bit "broader".

Say i have have slightly different prompts "what's 2+2?" vs. "can you please tell me what's 2 plus 2" or even "2+2=?" or "2+2" for most applications it would be useful if they all produce the same result

The form of the question determines the form of the outcome, even if the answer is the same. Asking the same question in a different way should result in the adherence to the form of the question.

2+2 is 4

2 plus 2 is 4

4=2+2

4

Having the LLM pass the input to a tool (python) will result in deterministic output.

- write formulas for me (single cell but let me say in english what I want then give me the formula) - do more complex things like "add another week" (1) insert new column (2) copy all formulas (3) adjust the fixed date - Explain a sheet / formula / section - Help me understand where this value comes from - do large scale formatting adjustments/alignments - find discrepancies / inconsistencies

Your requirement is to be able to automatically add rows (add corresponding formulas to calculate the corresponding data) or filter data through AI, which may allow AI to create a new data (automatically generate corresponding rows and make the data simpler)

Interesting. I've not used it but this sounds like the tools that were exposed to it need improvement. Everything you describe seems like it should be mostly in reach of current larger models.

@OP super inspiring. I'm working on a universal capture SDK, a bit like rewind.ai that would make it easy to grab information from screen and then store as Markdown etc. Have you ever wished for something like that?

Super cool to see this just now. We are building in the space of computer screen analysis and started to experience something similar, hence want to build something similar for Pixels instead of voice.

Would love to chat to you, jan@kontext21.com

Thanks, Jan! Would love to hear more about what you’re working on! Just sent you an email.

which model do you use locally?

Cool idea. How does it work. Do you expect me to run it on "unorganized" folder and afterwards it's better organized?

Just today I was wondering when I can just "ask for the file" and get it faster than navigating folders.

I'm wondering what you think are the advantages of putting them in folders?

You "can ask for the file" by searching for it, but it will necessarily take some time. Another approach would be to maintain and always update a database of files with locations. But it's still a good idea to keep files organized, such that each item has its own place.

For me it's a bit like typo in phishing/scam emails. If you think "oh my god this scam is so bad how can anyone fall for it" then you are probably not the target group.