tkdb

How much aesthetic freedom are you giving to opus?

The technical ability of anthropic models continue to steadily improve, but I still find the creative and compositional choices to fall far short of anything I would ever use.

One of the better TLDR's I've seen recently.

I'm not sure consumer rights protections would deliver better experiences.

Government is predictably slow, inefficient, and does just about everything at a service level that is only accepted because there is no other choice.

We are seeing an increasingly exaggerated cycle that goes like this: 1. Innovation takes huge investment to launch 2. Early adopters enjoy amazing new product(s) while companies bear losses 3. Companies eventually must face debt so quality of products decrease and price increases

OpenAI and Anthropic have perhaps set the record for going from 1 to 2 in the biggest (financially) and fastest manner ever, and we are all hurtling towards step 3 with a possible too-big-to-fail scenario.

Back to Disney Plus: I don't have a sagely answer. Free markets do sometimes need government to curb monopoly power, and Disney is certainly has a sort of monopolistic power due to their huge portfolio of franchises and content.

This essay format is surprisingly challenging, and one that is fresh every time you try it.

One that stuck with me from the NPR series is "I believe in BBQ"

Give it a try!

Agreed.

I think there is another overlapping factor at play too: good writing is novel in some way. AI can produce so. much. volume. so even if some of the patterns it uses have been markers of "good" writing, those same patterns become almost unbearable when you have millions of people pumping out content using them.

I wonder if model router adoption is higher for LLMs performing in-app functions, instead of the author's case of writing software?

One of the ways I can make a custom app/harness for a given task more of a competitive advantage is by striving for model agnosticism - the thing works and so has the valuable part outside of a model.

We've already seen older models being unavailable, and the newer (supposedly better) model doesn't do the work as well for some reason. Or a service is just down for minutes or hours.

Unpopular take: AI slop is actually just slop. (Most) people ultimately don't or won't care if writing or other media is partially or wholly authored by AI. Same with art. I feel as much righteous indignation as the typical person as I write that, I think. "I want humans making art that makes me feel these profound feelings!" I say to myself. But if I don't know who the creator is, all I have to go on is the art.

A sort of related concept is that I may dislike, actively hate, or be completely indifferent to some of the authors and artists whose work I love. I love their work, but I may not love them. I may choose to spend time with people whose art I dislike. So it is with AI. I dislike the idea of AI, so I claim to dislike the art.

As amazing as AI writing is, we are still in the uncanny valley, and that is why we hate it.

Imagine the holodecks in Star Trek, which for my argument I'll consider to be significantly more advanced AI, and significantly closer to crawling out of the uncanny valley... the people using them ultimately hit the same wall of "oh but this falls short of real human experience."

Is this arguing that, in order to produce commercially valuable music, all music is conforming to a set of standards, or traits and conform to "a grid"? That's how I read it.

Does this (2019) article propose the opposite? That hit music usually contains a sort of pleasing surprise, and that over time our hits have to become more and more unpredictable musically?

https://www.independent.co.uk/news/science/pop-music-good-br...

That's now I'm reading it. But I'm a music consumer, not a music creator, so I may be missing some nuance or conflating related ideas.

...and LLMs seem to be another obvious threshold, right? Perhaps the fastest and most dramatic introduction ever of new, automated ways of producing the same output?

>These findings have implications for species today

Mainly, crocodiles may survive a mass extinction event caused by humans.

I agree! And I see it as a continuum, not a single or distinct threshold.

At some point, it was clear to everyone that punchcards were no longer the most efficient way to program. People who chose to keep using them could for whatever reason could be considered artisans.

I also thing we should avoid automatically attaching to "artisan" value. I know some artisan bakers whose bread is amazing and sells out virtually every day they bake (alas, they still don't make much money which is why most people don't eat artisan bread). So artisan products can be better than their modern counterparts, but it's a mistake to assert it automatically is higher value.

Someone who writes code without LLMs is now an artisan software developer.

An artisan is (by definition) someone who creates manually, using traditional methods. Before there are automated or higher tech means of creating the same output, artisans don't exist, they are just craftspeople or tradespeople.

A factory worker, or factory owner, can care deeply about the quality of the product, and can be "in the loop" as you write about, but they are in the loop with the factory equipment vs. in the loop in manually crafting the product.

I don't think I'm refuting or in any real disagreement with you, in fact I may be here pushing artisan pedantry because I didn't see the literal use of the term coming through strongly enough.

It's amazing how much of AI workflow advancement collapses down to "now you don't have to copy/paste!"

I mean... the battle for books rages on in schools, from K to University. AI has been a boon for the book side (schools, universities, even whole countries banning not just AI but laptops/tech in totality from classrooms).

If you long for books for your children, join the fight!

I also hope to see a rebound for books and reading for all ages.

Honest mistake.

Those, those are ~thoughtful~ icons and not at all default.

Books will soon be obsolete in the public schools… Our school system will be completely changed inside of ten years.

– Thomas Edison, The New York Dramatic Mirror, 1913

What a quote! I love these takes because it instills a sense of skepticism for what our (wise?!) technologists are saying today. Especially the ones who are making the technology. No one knows how it will play out. Books remain.

Regarding your son: I am critical of adults' ability to orchestrate effective learning for children. Err on the side of honesty and directness, even in situations that are personally uncomfortable.

Most children are remarkably smarter and more capable of understanding than we (adults) give them credit for. Your son will likely have a better memory than you in the years when the acquired information disparity is highest (you having a lot, him having very little). Children learn early on how and when adults are being untruthful or partially truthful.

You cannot read too much to your son. You cannot talk too much with your son. You cannot listen too much to your son.

I want to understand this more - are you saying "would pay for itself" if you are offering chatbot commercially? Or using it yourself in lieu of ChatGPT etc..?

This is not to rebut your point, but I think for a lot of people/teams the silicon valley "lose money until we make money" pricing from OpenAI and Anthropic will continue to win (assuming the numbers below aren't completely off).

I'm out of my league on knowing if these ChatGPT projections are close, a little off, or way off, but we currently pay about $5k per month for Claude for our whole team. The projection for "run Deepseek on AWS to replace Claude" is coming in at $30k+ for GPU hours.

I have an ipad pro. 90% of the time it's a 3rd monitor for my macbook at my desk. (Very expensive per pixel for that use case!!)

It's nice to have for movies on flights.

A few times a year I use it as a portable second monitor.

I use the stylus for sketching sometimes.

Being self-critical, I overpaid for something I thought I would use more.

I'm not a fan of and don't use the ipad OS features that basically make it more like a mac desktop.

I should clarify: I have never attempted to replace by computer with an ipad. It wouldn't work for me.

Those who I have observed who have done so basically use the ipad for email and writing.

I was responding to "why would anyone use an ipad ever for anything?"

I mean - pre google had the problem of "these two webpages are both about how to take care of your dog, but which one is higher quality?"

Backlink ranking was/is a reliable heuristic to answer that question across virtually all topics.

So I disagree that the "problem was not about quality."

Spam/scammers will always be in whatever landscape there is, and I agree part of what google does it keep them at bay.

Yes, Keynote is (was!) my go to. But it's been years since I've been presenter guy so I may end up with a vibe-presented custom solution if I did it today.

I've used and like https://github.com/nyblnet/bento, but more and more I'm just going to self-contained html because the effort in to get what I want is just so low.

Let's imagine that virtually any software can be reverse engineered with enough tokens.

The bottleneck is the human effort to deploy the tokens.

I imagine that bottleneck is much, much, less of a constraint for openAI, anthropic, and the ilk.

LLMs are facing the same problem pre-google humans faced:

Is this information on this website true/high quality?

We need a google for LLMs!

Oh, but LLMs don't consume ads.

We need ads for LLMs!

Help.

This is a myopic take.

I've found iPads to be fantastic presentation tools.

Having an ipad on a conference/convention booth is amazing - airplay my rotating content to a big screen, and when someone walks up I can grab the ipad to delve into details based on the conversation.

Early on (first gen ipad) I had a whole client team nickname me "the ipad guy" because I showed up and presented across a few ipad apps where I could pinch, zoom, and annotate live instead of showing a vanilla PPT file.

This is already well underway with humans on both sides - especially in appliance repair (which is different but a similar technical/commercial model).

I have personally experienced several warranty calls for major appliances under warranty. The "tech" who shows up knows less about the appliance than I do, and the first thing they do is get on the phone with their own support call, who attempts to figure out what the issue is.

It's maddening for the consumer (me) because it (again, in my few experiences) doesn't produce a quality resolution quickly.

An actual appliance tech who still does service calls is like a unicorn for me - once I find one the are getting all my calls. My current guy for washers/driers tells me he is the remote tech for a few major brands, and spends a lot of his time each day on the phone with the onsite meat puppets.

This is interesting, and similar to something (I don't recall where, I think a book by Gladwell or similar ilk) I read that was basically: "Doctors with great bedside manners get sued way less than Doctors with poor bedside manners, even if their track records are about the same."

Right. I think there are two cases here:

Legitimate idiots who somehow acquire professional titles/positions. Things like "Florida surgeon removes wrong organ."

Then you can have honest mistakes by otherwise good practitioners.

What I want from my doctors and dentists is a sort of consultative collaboration, but the systems (in the US) are optimizing for transactions.

This is horrifying.

This is a lesson we should teach early and often:

Doctors, Lawyers, Dentists, CEOs can be idiots and may not know what they are doing. I hope the last vestiges of unquestioned authority and decision making in medical settings will be gone for the next generation.