Teams and Copilot didn't exist 25 years ago. It's the fact that they bundle a whole office stack under one subscription, which means companies end up being 100% Microsoft for cloud storage, messaging, email, AI, office and so on is the issue.

Just Windows and MS Office being incompatible with anything else apparently didn't meet the threshold.

I honestly can't understand how Teams can be so very, very shitty. Microsoft must use it internally, and most of the staff must hate it as much as we all do. Yet nobody at Microsoft is able to fix it. In fact, it keeps getting worse.

One aspect of enshittification that people don't talk about much is that sites do need to make money to continue. They can't be free forever. Often they make money in really shitty ways, like the recent RTINGS debacle with subscriptions, but you are absolutely right that none of them can continue to provide a free service forever when servers and staff cost money.

It's impossible to know how worthwhile a $2000/year saving (using one of the examples in the summary) is. They say their audience grew by 22% in a year, but not how much they expected it to grow had they stuck with Substack.

While you could do all this manually, AI combines some tools in ways that makes it a hell of a lot easier.

Notebooks are a good example. You can OCR them, you can type stuff in manually, you can extract all the text and throw it into a password cracker. It's laborious though. Now you can just take a photo of every page on your phone, throw that and your personal files from around that time, and have the AI do it. Maybe throw in some ebooks you facebooked, which you read around that time, in case you used a phrase from one of those. Some song lyrics too.

I've been doing something like that with AI lately. Take some order sheets and photos of bags of electronic components, and have the AI extract all the information and maintain a spreadsheet of my inventory.

Well, FWIW, I've never found a site *through* substack. They were always referrals from something else I was reading, which might have been *on* substack.

But that answer is false.

Your post was quite reasonable, and probably true, until you wrote "AIs aren't capable of reasoning". There *are* definitions of reasoning for which that it true, but they aren't the ones in common use. Cicero would use that kind of definition in his "school of rhetoric", where he taught people how to win arguments". Socrates would not. He was trying to find truth.

Clearly AIs have limited reason. They can (at least in principle) do perfect logic, but the difference between that an reason is not well defined. (And logic can prove that you can't prove algebra to be self-consistent.) To me reason is evaluating a set of data and a goal, and using logic to plot a nearly-optimal path to achieve the goal. I think where AIs are generally most deficient is in their goals, though obviously they also have an imperfect understanding of the current state. (Well, so do people.)

That said, there are many areas where current AIs seem deficient when compared with people. This doesn't mean or imply that they don't have a modest amount of the features that they are deficient in, but merely that we expect them to have more. Think of capabilities as being gradients rather than boolean variables. This is commonly called "jagged capabilities". They're better at some things than most people are, and worse at other things than most people are.

How do you know?
I will grant that there are definitions of "feelings" that would make your statement true by definition, but I will guarantee that most people don't use those definitions.

If you want to claim "it's synthetic, therefore it can't be a feeling", you've deprived your mind of a tool for thinking in this space. Submarines don't swim, but airplanes fly. Perhaps it's not useful to think of submarines as swimming, and perhaps it's useful to think of airplanes as flying. And perhaps it's useful to think of LLMs as having feelings. (Also perhaps it isn't, but just asserting that isn't useful, you need to demonstate it. My wife found it useful to attribute feelings to her car. The model didn't work for me, but it worked for her.)

A quick search didn't provide an answer, but the indications were that the cosmic voids are much of the universe. The search of turned up things like https://link.springer.com/arti... , but didn't actually reveal the proportion of the volume of the universe that is contained within cosmic voids, but did *indicate* that it sure wasn't trivial.

Well, I'm no expert in the field, but that's what I've read.

"One time" is far from reassuring. Sometimes the AIs get it right. However, if I am sending an AI to it, it's too complex for me to figure out at first glance. I am typically sending it complex projects with a lot of steps to figure out. AI is a nice upgrade from Stack Overflow and a powerful tool. However, in order to justify the AI-washing layoffs, it has to be a lot more reliable than "one time." I get failures daily.

I have not been impressed with Claude's unit tests. They're usually stupidly verbose. I've thrown away entire batches of code when I see they take a simple function and start unit-testing the Java getters and setters...even worse, they don't clearly indicate they're doing so...and it looks like actual business logic until you look closely. I would love to have Claude write acceptable unit tests. That would save me time and help me make more robust releases. It just fails pretty reliably on those.

AI is not useless, we just have to be realistic. It's like self-driving cars. Someday I am sure they will be great, but they're not really great now. They are still an experiment to play with, not something changing our lives.

That's actually a very insightful interpretation.

That the universe is not uniform at the large scale is blatantly obvious. Just consider the cosmic voids. And that implies that "time" should be running faster within those voids (because general relativity).

This isn't generally dealt with by global theories because it's computationally intractable...but it's inherent in relativity. So unless you want to break relativity, areas with low mass have time running slower than areas with high mass. And the cosmic voids are HUGE.

OTOH, This is a different argument as to why the universe isn't globally uniform at the large scale. But it reaches the same conclusion (to that question).

Now anyone can throw the kernel source code, and any publicly submitted patches, at AI, the idea that you can just keep quiet about a vulnerability until everyone gets around to patching it is questionable at best. The chances of the same flaw being discovered in parallel have massively increased.

Big companies that run millions of servers can at least detect when vulnerabilities are being exploited in the wild, and delay disclosure until that point or until the patch is widely implemented. Not so easy for open source developers.

It's the realization that the old "many eyes make all bugs shallow" thing was never really true. Once code is working, people tend to ignore it. Only NSA types were doing proper security audits. Once AI tools became available to find bugs, this was inevitable.

Same thing happened with Firefox. Turned AI on it, found hundreds of bugs, many of the security related. The fact that only 3 people use Firefox now is probably all that saved it from being exploited earlier.