I like the format of this video except for that mid-plug for Patreon. Leave it for the end.

As to the actual text of the video:

At the end of my short adventure at MSFT I realised I organically obtained the abiity to read through Satya’s emails. He had the uncanny ability to take a single sentence and extend it into a 3-paragraph email. Like, he would send a message whose actual information content was “we’re cancelling the annual base salary increase” but it would take up your entire fucking screen. However, after receiving so many of those, after some time I was able to read it effectively – skip the first paragraph, there’s never anything of worth there; if the sentence starts as if it weren’t leading anywhere then it’s not, don’t bother; read every second word – there, you just saved 5 minutes and learnt exactly the same thing.

I never considered language ability to be any indicator of smarts. I’ve learnt your godawful language from scratch, literally anyone can do this, Elon Musk spoke reasonable English (before drugs and 4chan ate his brain). I don’t mean it as something virtuous, as if I was better by not having this “flaw”, but rather as… I never realised just how much of your image comes from that. So you’re telling me people think this guy is smart because he uses four-syllable words? Wow. One of the best engineers I’ve met speaks like B1 English, makes constant grammar mistakes, and speaks with an accent thicker than Yud’s skull, who the fuck cares, everything he says about software is pure gold. And now I realise he probably never got that promotion he was aiming at because some dipshit above him thought he sounded dumb?

There’s no wonder the managerial class loves genAI so much, their entire shtick depends on copious amounts of form hiding the roughly five words of substance they come up with monthly. At least Satya doesn’t have to spend so much time writing that slop I ignored anyway…

Very astute observation about how we’re a very language oriented species and that essentially AI is a perfect vehicle for confidence men. I’ve said before all LLMs have really done is create a computer that is as inconsistent as a human, and in a way, that works in the favor of these things as confidence trickster tools.

When scammers make their scams, they often aim them at the simplest people possible, and what folks like us see as obvious mistakes that give away the game immediately are actually there because they don’t want people like us wasting their time. They want the easiest of easy marks, and the easiest don’t know how to read so good.

An AI is in many ways the ultimate confidence man, always speaking in confidence and authority without understanding behind the words. Literally just empty words, like a real live conman.

But more importantly, like the scams that pull in the weakest, the mistakes AI makes may make it seem more human to them. This could potentially aid it as a tool for confidence men, both big (Sam Altman) and small (email scammer).

I personally am of the opinion that as a society we shouldn’t just being letting the most vulnerable being taken advantage of at all times, and we certainly shouldn’t be financially rewarding fools for making the tools to do so easily accessible.

I’m so upset that the common person is upset with the state of AI, but ARENT pushing for forced open source of it