hm-nah

Humans would be a lot better off, if we could agree on what we are progressing to.

When US has agenda of X and China had an agenda of Y, etc, we’ll be pulling at the common center of humanity.

The feeling is pervasive in the enterprise as well. Each one protecting and petting their own agendas…fucking humans.

I’ve had good success with:

1. Scan them (or take photos of each page) 2. put them files in a directory 3. Make a Python script that sends them to OpenAI GOT-4o 4. Store the text as a new file in the directory.

You have the find a good OCR tool that you can run locally on your hardware. RAG depends on your doc processing pipeline.

It’s not local, but the Azure Document Intelligence OCR service has a number of prebuilt models. The “prebuilt-read” model is $1.50/1k pages. Once you OCR your docs, you’ll have a JSON of all the text AND you get breakdowns by page/word/paragraph/tables/figures/alllll with bouding-boxes.

Forget the Lang/Llama/Chain-theory. You can do it all in vanilla Python.

Vanilla python

Mind your entropy.

Chilling

Anyone have insights working with Ikigai’s “Large Graphical Model” and how well it does on time-series? It’s proprietary, but I’m curious how well it performs.

Wasn’t this part of Teslas work, to use ground as a conductor?

IMHO, speech-to-text to LLMs is borderline spycraft already.

“Are you wearing a wire?” The answer doesn’t mean merely recording the conversation.

Now it could mean, a personalized “analyst” determining sentiment, correlating events/evidence/facts/(hallucinations)/building a prosecutors case…in real time!

Azure Doc Intel is good. It creates a semantic representation (is json) of the doc. The schema is useful with page, paragraph, table, etc objects. The bounding-box for each element is also useful. It flails with images, only giving you the bounding-box of where the image exists. It’s up to you to extract any images separately then figure out how to correlate. Overall I think it’s useful rather than rolling your own (at least for small scale).

Thank god! An example of human ingenuity going to clean up our human impact on the natural world INSTEAD of YAStory of boiling the ocean to generate the perfect cat meme.

There’s a difference between “finding dangerous info” in a public space (library) or via a mostly auditable space (the internet) and having “a friendly assistant to help you make a real mess of society” on an airgapped computer.

Really? Where?

Because it’s open source, Meta (nor other SOTA makers) cannot “recall” the model either. How many more chances will we get to get this right?

A jailbreak doesn’t “make a model do something actually bad”.

A jailbreak makes it trivial to “provide a human who wishes to do bad, the info needed to be successful”.

Depending on the severity of the info and the diligence of the human, by the time you “see evidence of a real threat”, you could be enjoying a nice sip of the tainted municipal water supply.

This ain’t a joke.

DemocracyNow.org. Listener supported, first-person guests, giving voice to largely silenced voices, daily show, 20+ years on air, syndicated on many local airwaves or via podcasts or video on their website. Pretty much the best news source on earth.

And Bob becomes a high value target from a single comment in an open forum. Woops

Apple will soon include A.I. in iPhones that will in essence provide an eternal simulation of each of us. Progeny will be able to interact with us in our own voices and likely in 3D. When openAI and its humanoid robots align, your AI clone will be uploaded to the unit. Unfortunately “you” won’t be able to enjoy or reciprocate “their” enjoyment. Similar to if you recorded a video to your family today and they watched it after you passed away.

I really want to explore this topic. Been taking daily work journal notes for over 3 years.

I don’t use specific note taking software; tried them all, settled on markdown files in VS Code + VS Code Tasks.

Tasks are bada$$.

Is there an archive link to the vid somewhere?

I’ve been thinking about the future of this type of “AI fatigue”. Microsoft 365 Copilot has embedded this feature in Outlook, Teams, Word, etc.

I’m concerned that people’s desire and ability to read content from coworkers will significantly decrease, because…why spend my life reading something that “Bob” didn’t even write?

I’m not the only one thinking about this. M365 Copilot even has a “Make it sound like me” button…

This is an excellent visualization. Would love to see Mutual TLS depicted.

Claude2 is advanced and doesn’t fall for the jailbreaks. Interesting.

I watched your awesome presentation. I don’t recall the topic of sandboxing code execution. AutoGPT has the option to execute code in a Docker Container. Curious what you think of the scenario where an LLM is “tricked” to execute malicious code in a trusted env.

> The problem is…

…multifaceted from my experience.

1. At the org-level…

  - no standard doc tooling
  - no easy-to-remember URLs
  - no simple permalinks
  - no easy-to-update “pages” (AKA markdown-based)
  - they certainly don't train employees how to create good docs
  - minimal if any doc standards

2. At the leadership-level…

  - they don’t incentivize good documentation
  - they don’t set the expectation that “everybody writes”
  - they don’t document themselves!

It’s no wonder docs are scattered, out of date and untrustworthy!

Hm…“…spy-tech firm…”

What are the tornado (or extreme weather in general) stats for other States (the world)?

https://bendauphinee.com/writing/ Link to the post

Low-code platforms like PowerApps and PowerAutomate are great LEARNING and PROTOTYPING tools.

Give an analyst with some technical acumen PowerApps and ask them to mock up requirements, and after awhile, they can produce a click-through prototype.

That same analyst will likely improve their UI, API, JSON and SQL skills in the meantime.

Low-code tools are also good for digitizing CRUD forms.

Much beyond that and things get very proprietary, very fast.

I’m also curious.