K0balt

23 ms·

361.
by K0balt 8mo ago
To me, accountability it the smallest part of the issue. If you are interested, you can check out what i have written on the subject here : https://open.substack.com/pub/ctsmyth/p/still-ours-to-lose
362.
by K0balt 8mo ago
Yes, you could, and while I believe this would be much safer (not at the pointy end of your stick, but safer for humans in general) when this deception finally made it into the training data it would create a rupture of trust between machin
363.
by K0balt 8mo ago
Ai in general is different not in degree but in kind to the current crop of language models.
364.
by K0balt 8mo ago
Of course not, but they act as if they do. Their inner life or lack thereof is irrelevant if it’s pointing a gun at your kid.
365.
by K0balt 8mo ago
Yes, they are easy to fool. That has nothing to do with them acting with “intention “ which is the risk here.
366.
by K0balt 8mo ago
I’m using anthropomorphic terms here because they are generally effective in describing LLM behavior. Of course they are not conscious beings, but It doesn’t matter if they understand or merely act as if they do. The epistemological context
367.
Still Ours To Lose – almost all LLMs share one remarkable behavioral trait(ctsmyth.substack.com)
3 points by K0balt 8mo ago|1 comments
368.
by K0balt 8mo ago
I felt this was relevant and worth discussing.
369.
by K0balt 8mo ago
Advanced AI that knowingly makes a decision to kill a human, with the full understanding of what that means, when it knows it is not actually in defense of life, is a very, very, very bad idea. Not because of some mythical superintelligence
370.
by K0balt 8mo ago
It’s definitely an issue when using coding assistants. If you are careful and specific you can keep things reasonable, but even when I am careful and do consolidattion / factoring passes, have rigid separation of concerns, etc I find t
371.
by K0balt 8mo ago
Hmm, isn’t it though? I mean, obviously there is a corporate policy issue here, but there is no way that bending models to suit military purposes doesn’t end up in the general training pool, especially since we use models to train models. W
372.
by K0balt 8mo ago
Ive been writing about this on my personal blog. IDK if its worth reading, but at least its not too long? https://open.substack.com/pub/ctsmyth/p/on-the-character-of-... https://open.substack.com&#
373.
by K0balt 8mo ago
lol. The best kind of legislation (rated by entertainment value) is always written by people with no real understanding of the subject being governed.
374.
by K0balt 8mo ago
I like the way you think.
375.
by K0balt 8mo ago
It’s not AI per se, but rather ai enabled robotics that can change the world in ways that are different in kind, not just degrees, to earlier changes. No other change has had the potential to generate value for capital without delivering an
376.
by K0balt 8mo ago
Actually I don’t think it matters whether or not you adopt it. Or resist it. At this point I don’t see turning this bus around. Which is although I’d prefer to slow things down, instead im trying to make the inevitable disaster slightly bet
377.
by K0balt 8mo ago
This is some hand wavey malarkey, basically saying machines can’t have a soul because of….feelings? Insofar as feelings are self-proclaimed sensations of discomfort or pleasure, models that aren’t specifically trained to say they don’t expe
378.
by K0balt 8mo ago
lol. See you in the food line in a decade.
379.
by K0balt 8mo ago
I, for one, welcome our new robot overlords.
380.
by K0balt 8mo ago
Finally!! I’m porting my whole codebase to cobol! I write SAAS suites for archeological sites.
381.
by K0balt 8mo ago
Unfortunately, studies undertaken by MIT over a decade ago show that when it comes to law writing and passing, voters have no statistically measurable input at the federal level. (Since citizens united) It’s all just identity politics. I wi
382.
by K0balt 8mo ago
I can’t see anything specific in the link, and you probably already solved this… but just fyi, if it’s the same unit I used, and you are using arduino libraries, the library that worked was called one-bit display IIRC, and it basically acte
383.
by K0balt 8mo ago
I wonder if this is a result of a Fourier transform type operation that turns the serial time domain into something that can be processed in parallel?
384.
by K0balt 8mo ago
Not all applications are chatbots. Many potential uses for LLMs/VLAMs are latency constrained.
385.
by K0balt 8mo ago
TTS, speech recognition, ocr/document parsing, Vision-language-action models, vehicle control, things like that do seem to be the ideal applications. Latency constraints limit the utility of larger models in many applications.
386.
by K0balt 8mo ago
I’m all for more intelligent cleaning robots. The object avoidance AI is pretty good these days, but some of the navigation algos are just total garbage, unable to deal with trivial anticipatable problems.
387.
by K0balt 8mo ago
For those who may not know what the claim is: That opus 4.6 can successfully complete a (cohesive single) task that takes a human 14.5 hours, 50 percent of the time. It is unclear to me if this is zero-shot or iteratively driven.
388.
by K0balt 8mo ago
A highly opinionated thermostat?
389.
by K0balt 8mo ago
That would be sweet. That the supermini type with the 0.46” display? Those are fun for lots of things.
390.
by K0balt 8mo ago
While I am sympathetic to OP for having been somewhat unfairly called out as prejudiced or even bigoted by an unprofessional and somewhat reckless AI, I think there are larger issues that we miss here at our peril. Ai systems, trained on hu

More ›