A computer that both does what you don't tell it to do and doesn't do what you tell it to do. I didn't think we could do it but - I tell you what - it's been done.

@lagrangeinterpolator @nightsky THRUTH .. NEVER USE AI 🚨☠🚨

You assume these people installing experimental non deterministic software on their computer would know how to purge a process (or, you know, not to hook up vibe coded slop to their inbox) but here we are. To get a director job in a big company, the main thing you need is an MBA, a willingness to do whatever the CEO asks of you and either a sociopathy or psychopathy diagnosis (sorry for the repetition, I know I already said MBA). Technical skills "nice to have"

Before they could ask grok how to stop a process it was already too late.

Not that it mattered as Groks advice to become the reichschancellor actually didnt fix this problem.

if you should ever happen to be short on resumes…

(it feels like a zero AI job board might be a good thing to have, but we’d need a way to vet submissions and handle anonymous submissions and inquiries so people don’t dox themselves)

Huh, not what I would have expected. I work for a company that has sadly shifted very AI-focused, with the exception of the actual engineers. Literally none of us likes or uses LLMs. Every other week someone from the C-suite reminds us that we are encouraged to use it, and get 300$ or some such in credits for AI tooling per month, and that they don't understand why it hasn't been claimed even once.

many have expectations to set up an ide day one and it start churning out patches

I just don't understand the thought process. They must realize that this level of automation wouldn't require anyone to hire them?

I really wish this was my average experience trying to land another software gig in the past year...

When I do code review these days, sometimes I genuinely can't tell whether I'm talking to the person or to the slop extruder. It often ends up with me repeating the same comment over and over again.

It seems clear that every single company that makes money off of software is or will soon be in a race to the bottom on software quality

A lot of younger people who are being conditioned to accept this stuff just weren't around to experience how unstable and unreliable the vast majority of PC software was in the 1990s, and a lot of more senior-level people must have willfully forgotten. I've been thinking about this more and more lately. The difference was that in the 90s, the major PC companies could port their enterprise-grade OSes with proper memory protection down to the consumer level, as hardware advanced and running a more complex OS kernel was no longer a big demand. Even then, it was an uphill battle, especially once you threw widespread networking and dubious internet-sourced malware into the equation.

End-user software has already seen a decline in quality and increase in user frustration during the cloud era, as many apps have become siloed blobs of JavaScript running on top of an extra copy of your web browser engine. I'm concerned that we're headed firmly back to the bad old days now, without the release valve of better underlying software stacks on the horizon. The main solution will likely be to rip a lot of this crap out and start over (which is already a pretty widespread approach anyway, my credit union is going on their 3rd online banking "upgrade" in 5 years). But that completely zeroes out the "productivity" gains, not that anyone touting such things will ever measure it that way. I suppose the cost of re-stabilizing the software industrial base will be counted as GDP gains instead.

Tge best part is that you'd still have to check if it did it right!

@lurker @nfultz > If it was any other defendant and a juror said “I hate that guy and he has no moral compass,” Broome argued, that juror would be dismissedI mean, maybe, but this is an Objectively Correct Opinion

Forget who said it (I think e.w. niedermeyer) but if you were a true Musk Hater you would lie your way into that jury no matter the cost

Like I ever stopped...

Link to the Zitron sneer

It's a pretty wild read. This isn't a rational doomer screed about the annihilation of life on earth, though it similarly bounces radically between being overly vague and overly specific to create the appearance of analyjsis and consideration and confuse when it's claiming a fact with when it's extrapolating a trend (hint: it's almost always the latter and the trend may or may not be real). Instead it's written firmly for the McKinsey set to convince them their bets on the AI future weren't dumb and actually it's the naysayers who will lose their jobs and homes. Also David might need to update his site because there's an offhanded reverse-pivot back into crypto.

I mean, sure, but it's still the CEO of XBOX on her second day on the job throwing her hat in the legendarily sus declining birthrates discourse in service of AI solutionism, it's not nothing.

It could produce the stupidest outcome though, where Claude finally manages to either destroy or leak the contents of (or both!) a business-critical system that nobody understands how to rebuild.

This feels really sad to read through on some level. So much desperation for connection with someone and willingness to take the psychic-style tricks in good faith as her messages get not only repeated back to her, but rephrased in an obsequiously helpful tone! but I can't deny the willingness to get chatty about configuration details, private APIs, and what's on the second monitor as soon as the coding assistant gets into flirting mode is hilarious.

Truly, the tech industry seeks to close the gap not by increasing the capabilities of AI but by diminishing the capabilities and richness of human thought. Good luck to all girlthings in these trying times, and remember that a doll still means more to someone than a MAU tally for Anthropic.

God I was just reading that and it's so hard but it's so funny because that poor girl freya seems to have caused a crisis for Kent by being genuinely enthusiastic about AI bullshit and making friends with chatbots.

I wonder if Kent is going to have to do conversion therapy on his AI girlfriend now. Ethically of course.

ProofOfConcept may not be sentient now, but once we figure out how to put programming socks on her, the +2 coding bonus will put her over the top.

computer, print awawa.

@BigMuffN69 Thomas Metzinger (“Being No-One: The Self-Model Theory of Subjectivity”, 2004) also argues representational capacity is required for consciousness, but in a much more principled manner and with many examples, at length. I notice they don’t reference that book, or indeed any book that’s younger than 75 years old. ¯_(ツ)_/¯

@BigMuffN69 @Amoeba_Girl My old thermostat ran a feedback loop with a bimetallic strip and a mercury switch. Was it sentient? It was certainly more reliable than an LLM.

@BigMuffN69 Cognitive science does not have a working definition for ‘intelligence’ or ‘sentience’. We can define ‘consciousness’ circularly in terms of GCS score, but that’s the same approach as defining intelligence in terms of its role in a Turing test.Anyone who claims to have a functional definition for any of these terms is trying to sell you something.When (if) this actually changes, it will be massive, significant world news.Until that time, it’s a useful metric for spotting con-artists and morons.

@BigMuffN69 @Amoeba_Girl I think a sentient system (1) has a memory of experience (2) uses that updating memory to color the signal from its sensorium on a partial order and (3) takes action seeking better outcomes on that partial order. Evolution would favor (3) aligning with being healthy, cautious, and successful at reproduction but that’s outside the definition.

Kolmogorov complexity:

So we should see some proper definitions and basic results on the Kolmogorov complexity, like in modern papers, right? We should at least see a Kt or a pKt thrown in there, right?

Understanding IS compression — extracting structure from data. Optimal compression is uncomputable. Understanding is therefore always provisional, always improvable, never verifiably complete. This kills “stochastic parrot” from a second independent direction: if LLMs were memorizing rather than understanding, they could not generalize to inputs not in their training data. But they do. Generalization to novel input IS compression — extracting structure, not regurgitating sequences.

Fuck!

Oh you should be extremely worried.

Yeah, they rebranded when they did the harebrained pivot to focus on cryptocurrencies.

I hadn't heard of square either. Are they the guys doing squarespace? No idea.

EDIT: Okay, I did hear of CashApp, and it goes without saying that you need an entire lock-in ecosystem and a crypto-gimmick around a fintech product these days.