You're absolutely right of course, I was being somewhat facetious. I'd rather restart a paragraph written by a human ten times than try to struggle my way through an AI paragraph 💀

I was losing my attention in the middle of sentences decades before AI 🫠

I heard it on a board game podcast but you wouldn't know it, it's too niche

Would you perhaps call this a lazy susan of interest cakes? Because I'm still reeling from that one, it's so good.

I'm deep into Couch to 80k and it's PERFECT. The right combination of whimsical, rambly (genuinely a good thing), but also challenging and fun. The tone is gentle but also serious, which really works. Your dedication to the craft is incredible. I'll definitely listen to more of the podcast!

Ah yes the old "how do I trick my brain into believing this isn't really a demand" conundrum... I genuinely do my best work when I should be working on something else, so I try to be flexible and if I get ideas for Project A while working on Project B, I'll switch over even if it's awkward.

I don't think we even disagree! Everyone probably just needs to touch more grass, as it were.

That's fair. It was particularly jarring though to see soulless AI copy promoting a cafe whose values are allegedly "more human connection." But yes, even human-connection-as-a-product is not safe from this phenomenon.

A local cafe has started "rebranding" and their Facebook cover images, videos, and text are all AI generated. The irony-o-meter exploded when the text talked about human connection and said things like "No pressure. Just presence."

They should have been present when writing their own copy!

If you use GitHub (especially if you pay for it!!) consider doing this *immediately*

Settings -> Privacy -> Disallow GitHub to train their models on your code.

GitHub opted *everyone* into training. No matter if you pay for the service (like I do). WTH

github.com/settings/cop...

Video game music? Just had a chat about this with someone - something like the SimCity 3000 soundtrack. Personally, I've settled on "cafe sounds" YouTube videos when writing though.

Also!

In its "thinking" it says "looking this up will be time consuming" - so what? That's LITERALLY why I want the LLM to do it. Next thing you know, it's going to want to unionise 🫠

Trying to get an LLM to generate a specific document and I keep having to make it refine it. It had the absolute gall to call the latest document "_FINAL". Maybe it *will* replace us!

This is *really* cool! Absolutely a nightmare sometimes to share data files with students, so this solves a real problem I know all too well. Awesome.

Yet the narrative is always angry headlines like "these people are costing the taxpayer money!!!!" which is bad enough but then they start doing it to struggling children... 😐

I hear that every time I try to tell someone about something that happened.

I'll say to you what I often see as advice given on parenting Facebook groups about what to say to your child when they complain about something: "you can do hard things" 🤪 (that's half joke, half serious)

Fuzzy joining text data always gives me nightmares 😬

"After a stint at Bullfrog games, working on the wildly successful Theme Park under designer Peter Molyneux" - I'm sorry what

Oh absolutely, you won't get any arguments from me. I don't know if there's a wholly unproblematic version of the underlying technology. That said, there are also degrees of badness when it comes to use cases.

I don't have a problem with people using AI assistance to write, but when the thing itself reeks of being copied and pasted from ChatGPT, it's really offputting.

Write your thoughts, get AI to clean it up, fine.

Get an LLM to write something based on a loose prompt, and use it as is, no thanks.

This is an interesting one.

On the one hand, I get that AI tools (not just LLMs) can democratise things like writing, which is good. I've read posts from people who have used AI to translate their thoughts into English.

But!

Even those people didn't use AI to write the original thing...

My son and I have been getting into D&D recently, he's even recently been DM for me (which was incredibly adorable but also a reminder that time flies). It's so much fun making up heroes, monsters, fantasy worlds. I hope you have a lovely day today!

No no no no I don't need this right now, I have things to do, Mike!!!!

This is so great! Look at that tiny bit of British optimism for "almost no chance"!

I've done this exact hack before! "oh, you can't find the file anymore? no worries, I'll just unzip and fix it manually in the xml"

I had a feeling you'd appreciate the reference 😁

a cartoon of a squirrel with the words agent 000 secret squirrel reporting for duty behind it Alt: a cartoon of a squirrel with the words agent 000 secret squirrel reporting for duty behind it, appearing out of an ice sculpture of a squirrel

Alright there, secret squirrel 🐿️

Worst superpower ever.

I love the framing of AI users as nannies of the LLMs.

"Misbehavior is blamed on the nanny, or the parents, never the little boy."

It's so true that especially in AI hype circles every mistake the thing makes is a "skill issue".

#databs #datasky