Recreating MicroSoft's Recall

[@deanbell5164]

The 1st thing that springs to mind is using this type of tool to monitor a security camera(s). I like it!

[@DV-ml4fm]

Thanks kris for an excellent video tutorial project.

[@DigitalMetal]

My pleasure!

[@user-td4pf6rr2t]

if you can't beat em, pwn them. 4:00 Lol I reference filmsbykris for mounting loopback too! You know you have video with a for loop where you echo base64 embedding into a cgi-bin server i think - skills like that in the ai space would probably push the tech heaps forwards especially multi modality. My favorite is still the generative text. Personally. Lols. A good video for the era - Bash Chat room/wall. || this was a good one.

[@patrickprucha5522]

very cool!!!!

[@mrussell8054]

Yeah, but how are you going to push my data to Microsoft without A.I.? 🤔

[@SzDavidHUN]

The thing I'd worry about is SQL injection. Imagine scrolling over a text like `; DROP TABLE history; --` or something similar, and now your database is empty :D

[@DigitalMetal]

that's really a non-issue. the web scripts are on loopback and only do read functions. you could also change the permissions on the file to where the web interface can only read and not write to the file. it's a separate script that writes to the file. also since it's a single file and shouldn't be very big you could do periodic backups of it to another Drive or another location on the same drive. I'm not claiming to be a database professional but sequel injections only happen when things are written incorrectly or permissions are incorrect. and simple thing to avoid.

[@youtube_gpt-i9g]

I highly suggest the AI developer scene. Especially if mixed feelings. Lack pf interest reflecting the core existence is all the more reason to be part of the growth. Bro - think of it like chat bot are getting upgrade and its not even AI. We just think it is.... You are like chat bot numero unoer docs. Im trying to get signal handles and granting permissions, Computation is intellect is nonsense. Narrow AIs will be AGI. Go small or go home. So like since losely compression stage 3 of 2 shot prompt message 0 and 1 only are important to context but third shot is always generally stable in regards of cognitive response even if you talk nonsense stage 0 & 1. Fun trivia. So even though the model doesnt want to add request for web scraper is dodgy handling the data is alway consistent pretty much always unless it says No in second step. So we can be talking about replacing api-key but plotted out similar data has scraped code at the end. If that made sense. see the community way needs literate folk! I want chatgpt be mean. So brainstorming when debugging. 12:52 even though chatgpt is bpe if corpus is too lengthy tends to lead the conversation at a certain point, 13:47 do you know anything about license to use. for custom weights? Like ChatGPT - look, best practices...

[@McGyver008]

Nice demonstration... but it 'misses' an important point. Because this looks very 'harmless', demonstrating, how "useful" it could be. But... first of all: Every browser has a very simple function built-in, called "history". Sometimes even that is deactivated... because of .... what? Much better it would have been to demonstrate, how easily this method can be used to create data ready for 'profiling' simply _ e v e r y o n e _ and... (because of the very versatile usability of personal profiles - even and especially it can be used heavily by advertising) how this can easily be used for manipulation of people in many ways. This is, what this crap was made for. Nobody (if this person isn't brain-dead) needs such a database to remember what (s)he has done. I can consult a browser history and... there also is - at least in Windows - and at that time even critizised heavily) a "Recent" folder. And for those, who _ a r e _ that memory-limited: Everyone has learned to write and read, so make notes - full stop. But I refuse to give a company like Microsoft such massive information. BTW... Google is - for very good reasons - for quite some time criticized ... because every company seems to be paid by Google to make them advertise for and develop specific Apps running on Android, in fact opening every kind of data processed to Google's analytics (think on online banking using a banking app...). Google knows many people much better than they know themselves.... what is quite concerning (at least, but hopefully not only for me)

[@cubernetes]

I highly disagree. Browser history is fundamentally flawed, since it's so superficial. How is it going to find that article that I read about reverse debugging that I visited 2 years ago? Problem 1: it only searches in title and url Problem 2: you switch browser, you clear your history, you switch operating systems. Do you really make the effort of importing and exporting your history? This is a universal interface Problem 3: webrot. Websites go offline every day, it's a huge issue. I made a comment on a video 7 years ago. I found it on my comment history of youtube. The video is now private and I have no idea what video it was. This recall feature could easily give me that context back. Now, if and ONLY if you can find a counter-argument to all of these 3 problem, i will reconsider my point of view, but until then, it doesn't "miss any point"

[@JNET_Reloaded]

good idea but cats really? nothing intelegant from cats dude you could of picked any good subject lol and ai is a must to have!

[@DigitalMetal]

The cat thing was a joke. Because It is cliche to watch cats videos on the Internet.