gnutrino

I've found LLMs to be terrible with ideation. I've been using GPT 5.x to come up with ideas and plot lines for a Dungeon World campaign I've been running.

I'm no fantasy author, and my prose leaves much to be desired. The stuff the LLM comes up with is so mind numbingly bland. I've given up on having it write descriptions of any characters or locations. I just use it for very general ideas and plot lines, and then come up with the rest of the details on the fly myself. The plot lines and ideas it comes up with are very generic and bland. I mainly do it just to save time, but I throw away 50% of the "ideas" because they make no sense or are really lame.

What i have found LLMs to be helpful with is writing up fun post-session recaps I share with the adventurers.

I recap in my own words what happened during the session, then have the LLM structure it into a "fun to read" narrative style. ChatGPT seems to prefer a Sanderson jokey tone, but I could probably tailor this.

Then I go through it, and tweak some of the boring / bland bits. The end result is really fun to read, and took 1/20th the time it would have taken me to write it all out myself. The LLM would have never been able to come up with the unique and fun story lines, but it is good at making an existing story have some narrative flare in a short amount of time.

So cool, this is what the web was meant to be.

Their AI review is sub par, but everything else is really good.

The Graphite review UI/UX is at least 3x better than GitHub, and also somehow loads faster. Same with the customizable PR inbox. Love it! Appreciate your work on the platform!

GitHub is slow bloatware at this point. I can get a PR to load at least 2-3x faster in Graphite than GitHub. I avoid going directly to their UI at all costs now.

They say money is the root of all evil, and I think that is the core issue. It's unchecked greed and blind nationalism. Political and racial polarization is profitable. Selling guns and ammo is profitable. Being a corrupt politician who helps their rich friends make more money is profitable.

I've always done a lot of the things listed in the post, but just considered it as part of the job. I like laying it out like this as a strategy. I think this is a great strategy for anyone starting a new job, entry level and up!

The “suggest change” feature in github is a great way to suggest nitpick fixes without coming off like a jerk. You actually do the work, and the author can easily merge in all those changes with one click. They’ll avoid making those small mistakes / style choices over time. Also it feels more collaborative and less “do this because i said so”.

The site the post mentions for the original code (https://reactive.network/hackathon) is an accessibility nightmare.

Yes, the payment processor will charge you for each chargeback. Stripe’s fee is $15 per chargeback.

The amount of fraud and scammers out there is insane. I worked on a platform that only had a few hundred in revenue a month (just starting out). We did many smaller transactions, and getting hit with disputes was a killer. If someone did 15 transactions, they could get hit with 15 chargebacks up to 3 months later. So for every transaction, even if it only generated $3 in revenue, the chargeback could be potentially $15. (And you lose the revenue!). So for one customer who only spent $45, you could lose $270.

Even when we knew the person was legit, and just wanted a refund, they would do disputes. We only won a handful of disputes. The bank / credit card company will almost always side with their customer, even when provided receipts / terms of service / conversations with the customer where they admit the product met their needs.

Will do, thank you.

Thanks, I'll read up on this.

My guess is that the systems are running image recognition models, and maybe OCR on images, and then just piping that data as tokens into an LLM. So you are only ever going to get results as good as existing images models with the results filtered through an LLM.

To me, this is only interesting if compared with results of image recognition models that can already answer these types of questions (if they even exist, I haven't looked).

Maybe the service is smart enough to look at the question, and then choose one or more models to process the image, but not sure as I can't find anything on their sites about how it works.

Yes you should feel responsible.

The iOS version for ipad is a great way to play it.

Have you ever been hired by a company you refused to do the coding interview for?

I had clear for a bit when I traveled for work more often. It was nice, and it's definitely not "for rich people". It's less than $200 a year. I bet most people pay more for streaming services for non-essential entertainment. I've always been torn on if it's something viable long term though. While they're at it, they should ban fast passes at amusement parks.

It's actually worse because you still have to go through the normal security check, you just get to skip the line. This may be dependent on the airport though.

Looks very similar to https://apse.io/, which uses OCR to build a searchable index of everything you've seen on screen. I like the open source aspect of windrecorder.

Hapland was a favorite of mine back in the day. I still search it up every few years! Good luck with your search.

"critical ignoring" love this term

Seth Godin did a whole Akimbo podcast that was written by ChatGPT, and the audio was AI generated. The voice was spot on, the content and delivery was dead. I almost fell asleep listening to it, which is NEVER the case for any other episode of Akimbo I've listened to.

We got Gabb watches for the younger kids, and probably will go with Gabb phone when they get older.

"Do not taunt Happy Fun Ball."

My original blog post received some interest here on HN. I turned it into a talk, which I gave at EMPEX NYC last month. It really seemed to resonate with the audience. What do you think?

Original Blog Post: https://blog.andyglassman.com/2022/08/write-admin-tools-from...

RIP Paul. This was really hard on mobile.

I'm still on 17.X, so that probably explains the discrepancy here. I think I tried doing a major version bump a while back and things didn't compile so I put it on the backlog and forgot about it. It may be time to revisit before I get too far behind.

I believe you are correct. I could probably remove the imports altogether, and just prefix those functions with `Phoenix.LiveView`.