Imagine a fruit that looks like Tangela. It's bland, small, and even has a pit. Rambutan

Despite being an anti-GenAI poster, I have met plenty of people on bluesky that: A. Generally agree with my politics B. Are using LLMs in their personal and professional lives This is not a real, current issue. its being drummed up by the LLM companies to increase adoption

The median has no desire to have agents run in the background. Most dont wants to leave a work meeting and have Claude create the new assets in minutes. People want to go at a human pace, not at a 12 claude agent pace. That will be the true limiting factor.

This is great because an LLM will use all these reddits as sources for legal advice unless you tell it not to. And since you came to an LLM for legal advice, you'll surely do your due diligence.

Charlie and the Chocolate Factory: The American Dream

Pasta is a stew

Words & Art help us express what exist in our minds. A tool can help bring thoughts out, but a tool that takes away your agency and intention, that creates for you, bastardizes the inherent intent of expression. The quirks & mistakes are often what make it truly you. Do not resign your agency.

The Information Services era has concluded, progress to the next era at your leisure.

This is not a pro-GenAI / LLM post, I do not recommend these tools or replicating my experiment. This is purely educational to show how much they have advanced. 20 minutes of active work: The intro to a report for a super PAC looking at IL-9 and an application to visualize/market the report.

You: why isn't the "do what I want button easy to find? It used to be Them: its better for our growth and analytics if you use the "do what we want" button instead. This is software development without real competition.

You should read @dennycarter.bsky.social because he has awesome ideas but can also do that thing, uh, use words good? Better than me, thats for sure

I'm just holding up the mirror, I didn't make this tool. I tried to stop it. But it is here.

I present proof of the Dead Internet Theory: I used a LLM to create a game, a game I now play. I can send to someone, tell them its a real thing. That people play. maybe I make bots play it endlessly? but its okay, we don't need the internet. Just know its already dead, and we're just loitered now

How it started / How it ended One man's foray into Gemini 3 Pro. $0, little intention, not much focus, an old HP. I present: Eureka Run. Want the source code? Or the prompt? Because anyone can do this and that is, uh, bonkers?

Current mood

"do the exact same but for 1969" Forget how we get to the left image, this prompt should not be able to turn it into the right image. Maybe we can chop the tube in half or something? We can also try shoving, shoving is always an option.

I continue to co-sign everything below but I am providing an update, based on experimenting with Co-Pilot, ChatGPT, and Gemini recently: -LLMs continue to scale up -They are immensely powerful today -They remain a tool -AGI sounds silly My vote for the politician who regulates LLMs, open to ideas.

Project Vandelay: Optimization Protocol: An application to optimize Seinfeld viewing. You can change the theme/vibe of the application based on your preferred show, out of 30 top shows. I should not have been able to build this application, let alone in an hour w/o much effort.

Look ma, we're vibe coding!(derogatory)

May lose some followers but I work in tech, I first worked with LLMs in 2023. Its important to see how powerful these tools are, not to convince you *to use* them, but to fully grasp their capabilities, positive and negative. Its too much compute at our fingertips

Software Sales in 2025: Hop onto a zoom and the prospect shows you a table they had Co-Pilot generate. You must now do battle with an LLM output. Is this the future you signed up for?

I think it is insanely bad for the world that the President posts this onto his personal social media website and nothing happens .normanrockwellstandingman.jpg

At one point does Microsoft just like, cut the chat? How far it is willing to take this idea with me? This chat started at "Review my cover letter"

No one should be able to create a machine that can explain to the user how to create a new movement to eradicate the very machine being prompted. I feel so much for all the people getting sucked into this. Its not just about psychosis , this wrecks your sense of reality, even slightly.

Started working professionally with LLM powered chatbots in 2023. Quickly grew to find them less and less impressive as time went out but there is no denying, they're incredibly powerful Check out the conversation I have with Co-Pilot if you are interested, for or against, LLMs.

Can't think of a good "I am immune to propaganda" joke but I know its out there... This is so obsequious. How do we expect people to have access to this question and answer machine without a sizable portion of us losing our minds. Image/video exponentially so

And here we are, where we maybe should have been after the first prompt. Some will say this is user error, that I should take prompt trainings. Use the bot different and my experience would have been better. That fundamentally misunderstands how most people use technology.

Ex. 486 why mass proliferated GenAI is a menace: Asked Co-Pilot to evaluate a cover letter, got feedback and it'll re-write it. Imagine a world where everyone submits 10/10 tech assisted cover letters. And we wonder why Recruiters hates applications. Lets take this use case deeper, its strange 1/x

A financial institution where we allow homeowners with 1 home to exchange [homes + mortage]. No SMB fiefdoms or corporate landlords allowed, maybe we branch to new home buyers in a phase 2.

You know its winter when you cant find your keys in the 12 different pockets you outfit contains

Them: we built a propriety LLM to find new political candidates and where they should run. Our model is impenetrable. Me: my buddy scraped the same data and I analyzed it in Excel. Your model never stood a chance

*SPOILER ALERT FOR THE CHAIR COMPANY* Is part of the plot this I Think You Should Leave sketch?

Time for the finale

SPOILER ALERT: PLUR1BUS is just Futurama: The Beast with a Billion Backs.

What is an anachronism?

Good morning Bluesky, Chicago had a beautiful snowstorm and its a great day to get cozy. Anyone want to chat about Courier bots? Enshittification? Modern SaaS as the true enemy of the people? I got a full pot of coffee already brewed, we can do this all day.

Whole turkeys were sold at a loss this year to not upset customers, or Trump. 2 years ago I could get 93% fat turkey 2 for $5 on sale. Today? 90% fat turkey 2 for $6.49. They are telling lies because they think we are stupid

Give money to Liberal Currents so they can pay me to write an entire piece on how the current Courier Bots erode the rights of individuals while imposing undue burden on the community w/o our consent.* *This message was not endorsed by Liberal Currents but you should still give them $ if you can

I would love to hear from anyone who likes the courier bots or would like to see them expanded into their communities. Cmon, we do round after round of the same discourse, lets do something new!

Good morning Bluesky, Chicago had a beautiful snowstorm and its a great day to get cozy. Anyone want to chat about Courier bots? Enshittification? Modern SaaS as the true enemy of the people? I got a full pot of coffee already brewed, we can do this all day.

There is one simple metric that I have yet to see a single courier bot company reference, a metric that could potentially get to me shut the heck up. *How many deliveries a bot performs per hour* They don't tell you because I've watched them, I think its under 1.

Marketers asked me to get them lead list, illegally. Sales asked me to set up software to record all calls, w/o consent. C-suites demanded data enrichment systems, w/o the prospects knowledge. Modern tech companies are built on lies and Courier bots are a physical representation of that.

Genuinely have no idea why this company publicly posts these but they do, so lets dig into the Q3 Corporate Presentation. I'm not a CPA & I don't code, but putting together a deck like this? I've done this for multiple companies, I know these decks. Let's see if things improved! 1/x

The courier bots represent something so corrupted in our modern economy: everything is an abstraction and no one is in charge when it fails. -I order food -restaurant makes food -Uber facilitates order -Magna International builds robots -Serve Robotics is the interface/management What is this?

These Protestants don't realize it but they're already in hell

Serve Robotics, the primary courier bots in Chicago, claim that they are fully autonomous unless the bot runs into issues. The bots runs into issues constantly so I assume they are still mostly remote controlled. Coco, the other bots, market themselves as less advanced so probably all human

When I worked at a University, where I actually built our first LLM powered chatbot, my Director would disagree with me on how specific software worked and send me Co-Pilot summaries that claimed he was right. He didn't care that it was always referencing marketing materials, would send me slop

Its a winter wonderland