LOL *NSFW* *The new Demote thread*

End of this week according to some rumour millings

 

Anthropic have claimed a revenue increase for $700m a month to $4bn a month since the beginning of the year. I'm sure they'll open up their books to prove this point any day now.

 

Sure, that’ll probably happen around the same time I start believing that annualised revenue is in any way indicative of actual revenue instead of just being a snapshot at any given point in time. So… never.

 

He'll do one of two things:-

1) Have his digital waifu hallucinate a better-looking set of books beforehand, or
B) Run crying to Dear Leader that Wall Street is being mean to him

 

What are your strengths?

I make quick decisions

...and your weakness?

I make bad decisions

 

Gareth Halfacree, is it still your firm belief, that the AI industry is heading for a bubble burst? There seems to be plenty of reasons to assume it is.

 

It’s funny, this claim is made quite a lot, often alongside other claims about how we won’t need data/database people gatekeeping access to insights any more. And yet here we are, multiple years into the “AI revolution”… and I still have a job.

The irony is that it’s actually more effort for me to prompt an LLM than it is to write a SQL query.

Natural language prompt: “I need to analyse data in My_Table, I need to know the average value of Column_A where the date in ColumnC is on or after the 1st of January this year”.
SQL query: “SELECT AVERAGE(COLUMN_A) FROM MY_TABLE WHERE COLUMN_B >= ‘2026-01-01’;”

I could do it even quicker in DuckDB if I could remember its weird syntax shortcuts off the top of my head. But SQL isn’t particularly complicated, and it’s been around long enough to make it trivial to pick up.

OK OK OK, fine, a mathematical mean is a bit of a noddy example, compared to sentiment analysis of natural language.

But guess what: that’s not terribly complicated either.

Give me a couple of hours to refresh my memory - it’s been a while since I did any of this stuff - and I can spin up a Python program to do this on far far larger datasets. You can run it locally, it executes in seconds, it uses free and open source software, it doesn’t need vast rooms stuffed full of billion-dollar computers, in fact it doesn’t even need a GPU at all, it doesn’t need you to plug directly into the power output of your friendly local nuclear power station, and it won’t cost you a single penny.

Maybe you’d get quicker results if you pay money to SatNad in order to use his slop factory to generate meaningless slop that’s devoid of meaning or value.

Or you could pay experts to have them share their expertise with you.

 

I tend to write in 'techgrish' i.e. "gimme sql select avg col_a from m_table where column_b after 1st jan 2026" and let the AI do the job of putting the syntax and format around it. But, to your actual point, it's basically a waste of time when you know the syntax anyway.

Yup, this. In my first job one of my first projects was prepping updates for Y2K I remember the crazy valuations and the memes about shares being printed on rolls of toilet paper, because TP was worth more

 

my gut feeling is when it goes it'll be dotcom implosion meets the 2008 financial crash... the only question becomes - When and/or who is the Freddie Mac/Northern Rock and is first to implode [probably OpenAI but who knows].

 

Yeah, and every investment firm is throwing in with the AI groundswell, including ones handling pensions. So, when it does pop, we all get another bite of the **** sandwich.

 

Thankfully my pension is all in SIPP and a few weeks back I put most of it in non-tech funds to insulate myself from it. It’ll still be affected some but ought to be a bit safer.

 

One thing that saddens me about all this AI boll@cks is how reasonable people seem to either not be aware of or just outright put their heads in the sand about Sam Altman being a misleading sociopath.

I was speaking to a relative yesterday about it all (all I'll say is that he's involved in AI but not from a technical perspective) and I was trying to convince to him that Sam Altman is showing EXACTLY the same behaviour as Elizabeth Holmes did before the collapse of Theranos. He believes "it's not the same".

 

150 years later and only the brand name changes

 

For the benefit of the record. I did check on the status of the pilots before posting it.

 

Not intended as any kind of oblique criticism . More that I was glad to see the safety systems performing flawlessly despite what looked to be a relatively low altitude.

It was only after I’d been gliding several times (I was in the air cadets) I learned that the parachutes we used were basically just theatre. If we’d actually had to bail out for any reason, we wouldn’t have had enough altitude for the parachute to open.

They were fine for an aero-tow launch, but that basically never happened. We always had a winch launch, and you didn’t get much more than about 1500 feet at most (IIRC).

Mind you, I also learned at the same time that the safety record of those gliders was absolutely impeccable, and even the suspicion that something didn’t look right was enough to ground an aircraft.

 

Today in 'AI is totally the future...'

https://techcrunch.com/2026/05/22/you-can-no-longer-google-the-word-disregard/

 

Just tried that, it’s absolutely true and, ridiculous…

EDIT: Just minutes later, it seems to have been fixed.

EDIT2: The above was on DuckDuckGo, tried again on Vivaldi and, it again does not work.

Both of the above on iPad.