When a human makes sure employees are being polite, they're reinforcing the social contract that comes with employment. When you remove the human from the equation it's literally dehumanizing. That's it. Thats the why.
So what's the aggregate perspective of the 99%? You've described the 1% well, but that's only... well to be honest it is probably quite a bit less than 1% of all humans.
Any thoughts? What do you think the average work-a-day Joe thinks about all this?
- Way more people think AI will have a negative effect on the US over the next 20 years than think it will have a positive effect (35% vs 17%)
- Even more people think increased usage of AI will personally harm them than benefit them (43% vs 24%)
- Women and men have a huge gap on this: 53% of men say increased use of AI makes them feel more excited than concerned, versus 30% of women (probably due to deepfakes, but also likely due to women being more likely to be progressive, and the big anti-AI memes going around progressive spaces).
- 64% of people think AI will net eliminate jobs over the next 20 years versus just 14% that think it will even just not make much of a difference
It's also worth noting that AI experts were wildly out of touch with those attitudes compared to the general population.
It does however suffer from the (maybe insurmountable?) problem that "AI" is an extremely vague term with many potential interpretations.
I expect the "AI experts" in that study may have had a different definition in mind than the general public.
I remain much more skeptical about the impact of AI image/video/audio generation on society than I do LLMs, but LLMs themselves have such a wide array of potential uses that their impact will vary wildly depending on what they're being used for.
LLMs are obviously a transformative technology, but I think as tech people we have to be deeply thoughtful about how we ensure humans can apply meaning through their work.
A lot of my peers not in tech are worried about their capacity to work. And work isn’t just about making money but finding meaning. LLMs if not used correctly can transform meaningful labor into some perverse form of consumption.
No. That isn't required at all. Fundamentally you lack understanding of how this happens. Yes, there is some port duplication. Yes it costs money. But it is not anywhere near as onerous as you assume.
> Fundamentally you lack understanding of how this happens. Yes, there is some port duplication. Yes it costs money. But it is not anywhere near as onerous as you assume
No, I understand networking hardware quite well actually. I'm also familiar with Room 641A. Room 641A did not capture 80% of internet traffic. If you think 80% of internet traffic could be routed through Room 641A you're not thinking about the infrastructure required to get it all there. It was a targeted operation on backbone lines that were right there.
While the most well known, there are other points of presence doing the same thing. Easy and trivial to duplicate traffic at line speed. It doesn't affect the traffic flow itself.
A decent number of people reading this probably do have secret clearance. But that's not really the relevant point.
Simply having secret clearance doesn't mean you can just go digging around arbitrary secret classified info that you have no business reading. And it certainly doesn't mean that discussion can be had on hackernews.
That is simplifying it to the point of a lab experiment. It’s a bit more complicated but yes, you can split light and route that light anywhere you want.
What you can't do is ship 80% of the traffic across the world to the US without either the ISPs agreeing, and thus a conspiracy of thousands of people in thousands of ISPs, or doing it outside the data centres, meaning millions of taps in various ducts around the globe, which would be found on a daily basis.
Edited to not be so flippant: I work in HFT/finance where recording all traffic is required I think by law and definitely for one's own sanity. We're able to maintain nanosecond trades while capturing ALL the traffic. It has zero impact on the traffic. This is normal, widely used tech. Think stuff like Ixia passive taps and/or Arista Metamako FPGA-based tap/mux devices.
I have. I have a background in high speed networking.
Have you ever paused for a moment to consider how much infrastructure would be required to send 80% of data on the internet across the country and into a single datacenter in Virginia?
If you've worked in HFT, you can probably at least start to imagine the scale we're talking about.
I have no data or information on the topic, but the use of English was fine for the apparent intended meaning:
"Almost 80% of communications go through a data center in X"
Does not mean that all traffic goes through a single data center in X. Just that it goes through one of potentially many data centers that happen to be in X.
You're right. It's fantastic to see how English comprehension is decaying, even in groups that supposedly are smarter than average. There's a fast decaying tendency in language comprehension overall, and I can only point to the fact that much of the new generation is unable and unwilling to read even a single book.
“Loudoun County currently has 199 data centers, with another 117 in development, according to Michael Turner, vice chair of the board of supervisors transportation and land use committee and Ashburn’s district supervisor.”
So every ISP in every country in the world is feeding thousands of terabits of data to a hostile American intelligence agency. Not just netflow, the actual contents, and not a single ISP employee has actually come out with this evidence.
I can believe IXPs in many countries will send netflow data to their state's intelligence org, but that's a long way from what was being suggested.
Another way to think about it, many websites the data gets transmitted before you hit submit, between various type ahead reactive frameworks, soft keyboards with networked spell checking, your AI powered mood ring, always listening smart watch/car/home etc. Grandad always said don't say anything on the radio you wouldn't say in public, well we're up to don't think out loud or see how your crazy idea looks in text before you edit the Mel Gibson tones out of it. Tinfoil hats are off, on, locked!
the time it takes for light to travel from los angeles to virginia is 12 - 16 ms, round trip is 30ms lets say - that is a noticeable delay, and it could be easily disproven that 80% of traffic is literally routed through VA
now.. could they just copy the traffic and send it to VA on a side channel? probably?
There is the small possibility that the NSA has found cracks in some of the popular cyphers and could actually make sense of the encrypted data. It's not completely out of the question, their cryptanalysis has been shown to be ahead of the public best efforts in the past. They demonstrated it back in the 70s with DES S-boxes hardening them against a technique no one publicly knew about until the 80s.
i used to work, 15 years ago, on a (permissive, not covert) monitoring service for a UK national public service, the NHS spine core. We used switches to mirror ports and capture traffic in promisciouse mode on a few dozen servers
split across a few datacentres that all the traffic went througg. We had certs installed to decode https. We could get enough hardware to do this step easily, but fast enough storage was an issue, we had 1 petabyte of usable storage across all sitesn that could hold a few days of content. We aimed to get this data filtered and forwarded into our central Splunk (seperate storage) and also into our bespoke dashboards within 60s. We often lagged...
You can only decode those https certificates if you are mitming them (and have a compromised certificate)
A copy of the certificate and private keys won't help thanks to the magic of Diffie–Hellman, you can't passively (assuming you haven't got a practical quantum computer) read the stream
Your company will have deployed root certificates to devices and run as a MITM. This is standard corporate firewall behaviour.
The point they were making was that you could tell via ping times if the traffic was literally being routed through VA unnecessarily because the extra unavoidable light speed delay that extra distance would add between a user and the server if they weren't already very near to VA. Could be mirrored via the type of monitoring you're talking about but that'd only get you mostly encrypted traffic unless the 90s cypherpunk paranoia turns out to have been true.
> but perhaps with more proactive action taken to free up resources by removal of masses who are no longer producing enough value to justify their expenses.
How bout we remove the heads of the wealthy and hoist them on petards outside their former dwellings instead? We don't need to justify the removal of these cancerous lumps of flesh we call "the elite". We don't need them. THEY NEED US.
I don't buy this at all. There's significantly more to management than tracking and hitting metrics. You have to actually do the hard part of interacting with people and understanding their needs and so on within the context of work.
I'm sorry, but vomiting up another bulleted list is NOT management. Its giving orders and then culling the % that don't comply. Wildly different things.
> You have to actually do the hard part of interacting with people and understanding their needs and so on within the context of work.
Which is why it’s surprising that management is salivating at the idea of replacing employees and expecting 1 employee to do the work of 10 with the help of AI agents. If the value you bring is to manage people, you shouldn’t be happy at the idea of fewer people having jobs.
Again, what exactly are their jobs? You can’t automate them but you won’t need them either. If an org shrinks from 1000 engineers to 500 or worse, do you really keep those managers/directors/VPs around?
> You have to actually do the hard part of interacting with people and understanding their needs and so on within the context of work.
How many human managers actually do that, though? How many websites performed satisfactorily before AI arrived? How well has technology matched what consumers really need or want? Maybe, as a society we have underperformed and nescient AI performs well enough (or even better) in comparison.
When over half the population have bullshit jobs putting them all out of work seems inadvisable. Doubly so given the current political powder-keg we currently exist in.
You are incorrect. There is another way to address this problem and I suspect it will come to this: average people will begin attempting to destroy data centers and their interconnection points.
Your trillion dollar investment to control the populace ain't worth shit when its on fire and the monkeys are hurling flaming shit at you.
Humanity needs to wake up to the fact that our supposed "leaders" only lead us toward servitude. Our economic and political systems are designed to keep the vast majority of people in either literal or figurative chains so 25 people can get rich.
The entire system needs to be smashed to bits for the good of the many. Because after all, the needs of the many outweigh the needs of a few, or the one. But not in current human societies -- currently we value cruelty and malice.
reply