HackerTrans
TopNewTrendsCommentsPastAskShowJobs

13415

716 karmajoined há 8 anos

comments

13415
·anteontem·discuss
Grok is not a serious AI, it's not suitable for professional work and has mediocre performance anyway.
13415
·há 6 dias·discuss
Very funny.
13415
·há 14 dias·discuss
I think there are some sockpuppet accounts active but what also contributes is that many people are absolutely fed up with US technological hegemony and welcome alternatives to core technologies from elsewhere.
13415
·há 19 dias·discuss
What validation databases and where can you get access to them?
13415
·há 19 dias·discuss
I'm glad I don't use Claude.
13415
·há 28 dias·discuss
I use copy & paste with a pro subscription. I guess I'm a bit behind in terms of tool use but it works great for me.
13415
·há 29 dias·discuss
It's the future of warfare and it will sink the inhibition threshold for starting wars of aggression further.
13415
·há 29 dias·discuss
Or Apple could just fix these issues.
13415
·mês passado·discuss
As long as you're human, you don't have a substantially different notion of trust than I do. It's part of human psychology. We had to evolve it as social beings. I've seen very ad hoc CS papers on "trust", though, which IMHO were based on nothing but fantasy. Perhaps you had one of those re-definitions in mind. I tried to contribute to this pointless literature myself for a while until I realized that it's completely fictitious. All you can say from a mathematical perspective is that trust builds up very slowly based on a numerous factors and, if betrayed, goes down very fast and stays there. However, the notion doesn't have enough substance for a fruitful non-psychological modeling; there is just not enough ideal rationality behind it.
13415
·mês passado·discuss
One thing is sure, pinning trust on trust chains down from Root Certificate Authorities is fundamentally incompatible with our notion of trust and an almost absurd idea to start with. Most people using a browser don't even know any person from such an organization nor would or should they have any rational reason to trust them.
13415
·mês passado·discuss
The problem is that the current trust model is totally untrustworthy.
13415
·mês passado·discuss
Personally, I think Reticulum is the parallel Internet. It could even replace the Internet Protocol, and whatever the IP protocol connects is in my view the Internet.
13415
·mês passado·discuss
> IOW if we don't know what consciousness is, how can we call LLMs conscious when we haven't seen even the barest intermediary steps towards our nascent concept of consciousness first?

We stipulate that other human beings are conscious from their behavior and how it relates to ours when it is accompanied by our personal introspective experience or "awareness" of being conscious during normal cognition. The process for ascribing consciousness to LLMs would be same: A stipulation on the basis of behavior that relates to our own behavior and how it appears to be linked to the introspective experience of being conscious.
13415
·há 2 meses·discuss
The biggest and maybe only relevant difference is that one of them works evidence-based and the other one doesn't.
13415
·há 2 meses·discuss
I don't disagree with much you've said, the only thing that bothered me after having worked for 20 years at university is the lack of realism of your views about scientific practice. Scientists have to deal with politics a lot, and if it's just with fundraising schemes that often explicitly demand social relevance of the research. That's particularly the reality of disciplines that are extremely relevant for society such as climate science. Of course, they need to assess to some extent whether the consequences of climate change are good or bad for society. This is a no-brainer, if you'd bother to stop and think about how they would structure and organize their research and get funding for it.

> A random scientist's warning holds no more weight than a hobo on the street's warning

You must be joking, and I wish people would do that less often when talking about serious topics.
13415
·há 2 meses·discuss
I really don't get you stance. Of course, you can make more fine-grained distinctions and that's fine. You can claim that medical doctors act as medical scientists when they conduct studies and as doctors (consultants) in their practice with patients. But that doesn't mean the value judgments aren't part of the science.

If a seismologist has evidence that an earthquake is likely to occur in a certain area, should they not warn the public about it? I would say they clearly should, and any other view about this seems bizarre to me. I find it equally implausible to not call a seismologist who warns about an impending earthquake a scientist. They're a geophysicist or geologist. Or take an astronomer warning about a possible collision of a meteor with Earth -- astronomy is a science, so why would that person not be called a scientist?

There is a an array of scientific disciplines for whom consulting (in your sense of the word) is a frequent, though not primary part of their activity, and we certainly still call them scientists. Material science, vulcanology, epidemiology, seismology, meteorology, biology, climate science, economics,... basically any science that involves the study of processes that might have important consequences for mankind.
13415
·há 2 meses·discuss
You're nitpicking. Medicine is concerned with what's good and bad for someone's health. Medical doctors literally advise their patients on that and evaluate the effects of actions with respect to what's good and what's bad for their health. What's good and bad for someone's health is simply one form of instrumental goodness. Other sciences evaluate in similar ways, though they are perhaps concerned with other aspects of what's good and bad. Climate scientists are not concerned with what's good and bad for mankind in some abstract philosophical way, but they should without a doubt lay out good or bad consequences of climate change. If the temperature sinks by 10 degrees Celsius in Northern Europe, that would be a bad consequence for the affected countries.

It's false and somewhat naive to claim that such evaluations play no role in science, they are a crucial part of many sciences. For instance, they're needed to find worthwhile subjects of study. Not everything is theoretical physics.
13415
·há 2 meses·discuss
So medicine is not a science because it's concerned with what's "good" and what's "bad" for someone's health? I find this kind of argument principally flawed.

Many sciences are concerned with the consequences of human actions and it's hard if not impossible to describe these in meaningful ways without applying some criteria for what outcomes are good (desirable, positively evaluated) and what outcomes are bad (not desirable, negatively evaluated).

Besides, there is a whole area of science that maybe is more like engineering but is clearly worthwhile, too, even if it's not strictly a natural science only. For example, urban planning might not be a science in the strict sense but it's clearly important and involves scientific studies.

If policy makers can't get from climate scientist's an evaluation of the potential consequences of climate changes, then who else would produce these for them? Should they just make it up on the fly?
13415
·há 3 meses·discuss
You're essentially saying "Don't trust Google at all and ask your local government to put pressure on Google" and I agree with that but you frame it in a needlessly apologist way. If a company makes a promise and breaks it, that should always be a reason for concern, and the article is right for pointing that out.
13415
·há 3 meses·discuss
Interesting, this seems to license forms of affirmative action that are unconstitutional in Germany.