iiinova — where your digital self keeps evolving

the agency and liability of the creators should not be hidden by the agency of the products, obviously

020

Did I miss something? They do avoid the language of agency regarding cops. That’s the whole exonerative tense business. ‘Officer involved shooting’ etc. will that lead to ‘AI involved Armageddon.’

120

OMG it’s fantastic! We tell everyone to watch it but no one does

010

EVERYONE TOOK THE BAIT. IT’S A BIT!! THE GUY IS OBVIOUSLY TROLLING. Lol. I mean, it’s a thing of beauty. The responses here are hilarious. He really wound everyone up.

040

Wavy Jones @postambition.bsky.social · 22/09/2026

It’s got to be a bit. It’s a parody.

120

Can you describe 'could not read' a bit more? Like very poor reading comprehension, or didn't know letters and words?

050

Using a chatbot for simple questions or text manipulation is not like using Claude Code or Codex to tackle a project. It is shocking how capable they are of doing complex projects in which they generate plans, correct themselves, work around obstacles, change course due to discoveries, etc.

000

LLM's are just part of the agent harnesses that are common now. The next word guessing thing is an outdated description that is comforting but misses most of what is actually happening. Part of what is driving this argument is that lots of people are stuck on the 'stochastic parrot' metaphor.

100

You mean that some people think we are not deterministic meat robots?

020

I'm sure it's getting worse but: "Boomerasking was first identified in a study co-authored by Michael Yeomans, associate professor of behavioural science at Imperial College." I think Jane Austen was well aware of 'boomerasking' and I doubt she was a pioneer . . .

111

“Sort of ok code” that solves your very specific problem that you, a non programmer but domain expert, can make yourself is immensely valuable

000

Commercial apps are full of features you mostly don’t need made for other users.

100

Wavy Jones @postambition.bsky.social · 19/09/2026

This is the big win. Not only that, but now lots of people who can’t code can make those custom solutions. The focus on if it can make scalable, commercial applications is missing the point.

100

There is something about $5,000 as a number. The vet industry has calculated that that is how much people will pay in a pet health crisis. I have heard so many stories like: "I took my pet to the vet and it cost $5k"

0140

Wavy Jones @postambition.bsky.social · 19/09/2026

During Hurricane Ian, "A storm surge reached unprecedented levels of 12 to 18 feet around the regions of Cape Coral and Fort Myers." www.earthdata.nasa.gov/dashboard/st...

earthdata.nasa.gov

VEDA Dashboard

Visualization, Exploration, and Data Analysis (VEDA)

041

Reposted by Wavy Jones

Emily Atkin @emorwee.bsky.social · 17/09/2026

A few weekends ago I went to Miami and attended 9 open houses in extremely high-risk flood areas to see what agents are telling prospective buyers about them. Here is my journey

heated.world

The open houses at the edge of disaster

What are buyers being told about climate risk in Miami? I posed as one to find out.

421280408

What you call it does not matter either, 'thinking', 'sentience', etc. Those debates are interesting for all kinds of philosophical reasons, but they don't add much to understanding the actual capabilities.

070

How does a mayor reduce wait times for primary care appointments?

000

The point is there is no game. All of the votes exist. The outcome is already fixed, it is just unknown until the count is done. You can count in any order, it doesn’t change the final count.

2775

I don’t care if the tools are really “thinking” I’m more interested in what they can actually do in practice. It’s just convenient to anthropomorphize.

100

Well I don’t think people pay with bills much. Like a tenner doesn’t have the same resonance when you are just waving your phone to pay.

000

The post is awful. The reply is a good example of how the tools accelerate certain kinds of work

210

I don't know how old you are, but in my lifetime I've seen many miraculous technological advancements transform what is possible--stuff that went from sci-fi to so mundane as to be used in children's toys. I don't get the point of taking the posture that this is just same old same old. It's not.

000

I don’t understand this point. You can use the tools yourself and see what they can do. You don’t have to rely on the marketing claims.

Belostoma's comment on "PI will not stop using Grok for revising *MY* papers"

OK here's a great description of concrete examples of profoundly useful stuff. Also, further down in the thread the poster makes the point that the debate of what you call it 'thinking' 'reasoning' etc. is beside the point and the point is the actual capabilities. www.reddit.com/r/PhD/commen...

reddit.com

Explore this conversation and more from the PhD community

000

Belostoma's comment on "PI will not stop using Grok for revising *MY* papers"

OK, I'm back with a concrete example. Of course it's an anonymous reddit poster, so grain of salt, etc. But this is a great example of the kind of stuff I am talking about. www.reddit.com/r/PhD/commen...

reddit.com

Explore this conversation and more from the PhD community

210

I agree, a lot of concrete examples in plain language would be way more useful than the debate over the meaning of 'think.'

The same place it comes from when you save money by buying something that is on sale.

010

And some of the individual things are kind of a big deal. It used to require real, hard won individual experience (and special aptitude) to do them and now the computer can do them with some simple instructions.

Well, it's not just information but skills and abilities. Yes, I suppose you could say that it is doing things that are known already. But it definitely a big new thing that a computer can not only do these things, but that it can on it's own figure out what to do and how to do it.

The systems have all kinds of flaws and are prone to hideous and terrible misuse and seem to reinforce malign tendencies in our economy and politics. But, I'm pretty sure that they can do stuff that is 'more than' what machines could do a few years ago.

010

OK, I have to think about that. I'm not convinced. The coordination itself is not something that machines could do a year or so ago. The individual tasks, maybe but not the coordination. Also, a year or two ago, the machines couldn't write software the way they can now--like literally could not.

210

Yeah, it's a real mishegas. Hard to see it clearly. In my own case though, I'm doing stuff that the systems are really good at and that I can't personally do. So for me it's basically miraculous. But it's also fraught with slop, errors, cat herding, lying, cheating, etc.

It's truly weird and astonishing to work with them.

Well the crazy thing about the agents is that they figure out how to do stuff, make up plans, spin off sub agents to tackle parts of the plan, coordinate the sub agents, pull together the results, figure out how to work around obstacles, find mistakes and correct them, ask for new input, etc. etc.

Hmm, I do stuff all the time with it that is completely outside of my personal abilities, stuff that I simply cannot and probably could never do.

I think it varies with the chatbot and the users. The users you ask aren't doing much. The so-called agentic AI harness things like Claude and Codex can conduct whole projects in startling ways. This is dramatically different from helping coordinate searches.

330

you can also ask claude to find the best evidence that contradicts your claims, FWIW.

000