walleeee

17 ms·

241.
by walleeee 4y ago
> That’s like saying every Tesla owner should be able to program their own car! Let’s open the source code of Tesla to every owner! Other critiques aside, these things are not very similar. You can't run someone over with a buggy re
242.
by walleeee 4y ago
the fact that it has encoded domain knowledge doesn't necessarily imply it "understood" its errors after you pointed out contradictions, or that it formed a coherent position by some process akin to reasoning at all times it
243.
by walleeee 4y ago
From the article: > “From our perspective, the big surprise is how little control we actually have. We have to play our cards very carefully and very wisely, or even that control will sort of evaporate,” Gackle said. “There’s often a str
244.
by walleeee 4y ago
It shouldn't be exclusively on the near term, but I'd argue human/machine collectives are a form of superintelligence with many of the same risks a misaligned AGI would hypothetically present us with Alignment problems aren&#
245.
by walleeee 4y ago
Yes, and this item... > AIs becoming sentient and causing harm for their own ends. ...tends to get a large attention share, while the others likely present much more immediate risks at least in the near term
246.
by walleeee 4y ago
I'm not sure I follow this answer. What are the entirely new paradigms? Writing is still the initial step. If text editing remains a core part of the workflow, why restrict the user's ability to edit code?
247.
by walleeee 4y ago
> All of them are solid, kind, high-quality people, the likes of which you are unlikely to meet in the regular world. This attitude may have something to do with the skepticism we low-quality people out in the regular world may harbor fo
248.
by walleeee 4y ago
> The problem is that "understanding that capitalism has problems" isn't equivalent to "having an actionable plan to solve capitalism This is a caricature and the same thing could be said of AI x-risk. There are plent
249.
by walleeee 4y ago
Thanks! Looks like good work. I hope this idea continues to get traction: > In some sense, we’re already living in a world of misaligned optimizers I understand this is an academic paper given to nuance and understatement, but for any dr
250.
by walleeee 4y ago
Until the alignment movement begins to take seriously the idea that we already have misaligned artificial general intelligences I think they are best viewed as a convenient foil Paperclip maximizers exist, they're made not only of co
251.
by walleeee 4y ago
Is there not an enormous difference between spreading gossip (in my mind, always about someone in particular, which this is absolutely not, or at least not to anyone not neck deep in bay area software circles) and relating instructive, an
252.
by walleeee 4y ago
It seems to me that if a civ with high technology is going to expend the effort to seed large numbers of solar systems with xenocide devices, it might as well spread itself through the same space instead
253.
by walleeee 4y ago
Right, this is not a new phenomenon. Whether it's meaningful to pin it on human nature, I'm not sure; we consider ourselves unique as a species largely for our ability to recursively adapt the environment to ourselves, and then ou
254.
by walleeee 4y ago
Indeed, and someone with money and a desire to prevent actual class conflict can get a lot of mileage out of encouraging tensions in the narrow, purely expressive sense you describe
255.
by walleeee 4y ago
Some people argue we might consider abolishing (or at least ignoring) institutionalized peer review, e.g. Liam Kofi Bright. The blog post/Arxiv model seems to work reasonably well with no fees or formal referees, and was more or less h
256.
by walleeee 4y ago
and funds a good deal of biological research in academia
257.
by walleeee 4y ago
> The technology needed is well-established, just collect a lot of vials of water, soil, stream bed cores etc. on a grid scale focusing on the entire region, paying attention to downwind zones from the accident in particular. Then they n
258.
by walleeee 4y ago
Clearly true, if you value money more than whatever it is you founded your startup to do. Which is likely true, if you're in the business of selling startups. Now we're going in circles.
259.
by walleeee 4y ago
One wants them to be money obsessed only if one thinks selling startups for money is a valuable economic activity
260.
by walleeee 4y ago
Do you imagine a Mars colony will not have its own politics? Politics is arguably motivated in large part by scarcity. On Mars resources are a lot harder to come by than on Earth. You could make an argument that a shared struggle against ex
261.
by walleeee 4y ago
> Most plausible post-apocalyptic Earths would be far easier to live on than Mars. thank you how are we supposed to build a second home on a dead, ruthlessly hostile planet until we demonstrate ourselves capable of stabilizing the biosph
262.
by walleeee 4y ago
shorter time horizon could also be related to pervasive awareness of existential risks going unaddressed maybe people also (in many cases correctly) intuit that any effort they put into their role in society, much less extra effort to help
263.
by walleeee 4y ago
> technology does not drive history, people drive history this fails to grasp the changing nature of agency in a world brimming with hybrid intelligence we would perhaps like to think we and we alone "drive" history, and I agre
264.
by walleeee 4y ago
For me, it's helpful to think futures as attractors. We are their agents: futures pull themselves into being not only in blind compliance with physical law but also through us . Which will we welcome into our minds and, through this,
265.
by walleeee 4y ago
> There is no practical way we can achieve the goal of survival through reduction of CO2 emissions. It's a farce It's a farce because it's already too late for emissions reductions to cut it. We need to pull carbon out o
266.
by walleeee 4y ago
> I'm sick of being lied to and cheated by clowns Then don't work for them. Don't apply to startups building yet another ecommerce site. You can find a job writing code for something that won't evaporate next year, th
267.
by walleeee 4y ago
> The fact is we got to all the modern conveniences by the path of least resistance This is pithy and raises a broader question: once we have started down the path of least resistance and identified (or already tripped over) hazards on i
268.
by walleeee 4y ago
Or hydrology
269.
by walleeee 4y ago
an 80k starting salary is hardly grim and if you happen to become a decent scientist or technician along the way there is usually work to be found, but yes, the phd mill incentive matrix is a blight and demonstrates broader misalignments in
270.
by walleeee 4y ago
the canonical brand name university experience in the US these days can hardly be called independent living a growing portion of the bloat discussed in this thread goes toward increasingly luxurious lodging, amenities and services generally

More ›