I remember reading that previous dose of hatorade in February. Personally I thought it was too lone-man-crying-in-desert to share here. Not that I disagree with his facts, just that I wasn't sure what to make of it. There is a blindness to his hatorade which is making me suspicious.

Still, if he ends up being right about OpenAI's terrible position, does that inevitably mean they're screwed? Are they for sure on a one way ticket to becoming the next Lehman? The next Elizabeth Holmes? Or are they simply running in the red like Musk has done many times before, banking on OpenAI to coast on delusions for long enough to figure out how to lock people into their tool ecosystem as they seem to be doing?

(sidenote: I tried to create an Elizabeth Holmes-Forbes-cover with Saltman's face but the GPUs are melting again over at OpenAI and it refused to do so due to internal errors and made up a BS reason for why. that bout sums it up...)

The release of 4.5 was the sign on the wall, for sure. It was the largest model training they could have possible done, the most horsepower put into that part of the equation and out came nothing. Nothing new, nothing significantly better, just a ginormous model that is supposedly eye-wateringly expensive to run. OpenAI tried their absolute best to cover it up with a big "...anyways" sign but it's a grand failure whatever way you look at it.

My money is not on OpenAI, just to be clear. But I would also not be surprised if it takes a while for them to deflate. I personally believe 2025 is the year they will (start to) get pwned by Google, if they aren't already:

HYPOTHESIS: AI Boosterism is the consequence of despairing males looking for Teddy Ruxpin to replace their need for human contact

Can we add the Unethical Zurichians to your observations, just for the hell of it? If I remember correctly, their (now removed) early abstract claimed that by consent-free scraping of users' Reddit accounts, they could like 3x their ability to convince users of different viewpoints.

Let's draw a quick old-man-yells-at-cloud parallel. Millenials like me got phones in early high school, smartphones in late high school/early college. We got addicted to them but know what it felt like without 'em so we just feel bad about our own discipline. Zoomers got hit hard by both smartphones and social media in formative years. They got addicted to them, vaguely know better but their attention span and social skills got so rekt they're locked in and depressed. Gen Alpha grew up with smartphones and like 40% of 2-year olds now have an iPad. It is like fish in water. I see mostly Millenials now pushing for smartphone-free upbringing.

I think there's a same parallel now with AI (boosterism), but one generation down. Zoomers know the Before Times and they dislike it but they do use it. Alpha is gonna get hit hardest, is my fear, with high school cheats at the ready but without any kind of checks and balances in place from schools as those are still being figured out.

Socially, I think we need to move real fast to figure out guardrails or norms or anything before we get a perpetual Her situation for any mildly loner kid. I am a bit scared at what access to a friendly, conversational, supportive AI would've done to a younger, much lonelier version of myself. My Ruxpin was spending a lot of time on forums, but those had moderators and some mores...LLMs have none. With memory improvements, with the Unethical Zurichians showing that a personalized AI is very good at manipulating people, there is a real risk of LLMs seriously shaping people's social skills. I mean, Character.AI already has like 250M users? And that's just fringe horny users specifically seeking it out.

Maybe I should reread Reclaiming Conversation,