Often the problem best solved by AI isn't "How do I use Anisble to do X" But simply knowing that you should be using Ansible at all. Or which features to use. Then you can be laser-focused on learning exactly the right part of Ansible.
It also helps you to move away from familiar but hacky solutions towards much more standard and robust ones - the AI doesn't approach the problem with your many years of battle scars and baggage but instead sees your problem space with fresh eyes and recent data.
But why use AI for that in the first place? You can just, uh, google things. Like "software that does X". You would get a better result from forums and blogs and such. When doing research, AI really feels like an unnecessary middleman when you know how to do research on the internet. Which I would argue is a basic skill in the modern world.
So your point went from: they shouldn’t have done what they did because it likely took much longer than necessary by using AI - to: they shouldn’t have done what they did because they didn’t take long enough learning it without AI to do it. I think you moved the goalpost to the opposite side of the field.
In this thread you chastised them for “using it in production” when their use case was simply a lab for teaching. And you suggest that googling for days and reading countless blog posts and forums is impervious to wrong or dated information, which is quite a position to take.
Their use case was a perfect fit for a LLM, and I suspect it saved them weeks of effort in the process, with likely fewer security flaws as a result. But keep up that anti-LLM advocacy, I’m sure you are very close to achieving the goal of society abandoning the technology.
Keyword based search (Google) requires the user to know which keywords to enter. It's very bad at finding things related but with different words to what the user is trying to search. Sometimes half the challenge is figuring out what the experts call the task you are trying to achieve.
Set up some desktops with code -> Declarative configuration management -> Ansible
This task seems easy to an expert. As someone who's been a professional learner (as a programmer) for 25 years, it's far far easier with ChatGPT to get a summary of a field and a starting point than it ever was with just Google and some random blogs bouncing around the internet. Now with modern o3 (Chat GPT Pro) models you can go from a vague question to a summary of suitable solutions and some code samples in one prompt.
Now obviously AI is not able to create complete solutions alone. It's just another tool. Sometimes ChatGPT makes glaring mistakes in how it approaches a problem. So (for now) it is still a collaboration requiring smart humans in the loop.
Could a single-stage-to-orbit spaceship, something that could operate rather like an aeroplane, be built with just rocket engines? Well, actually, yes. In the 1980s, NASA and the US Air Force spent about $2 billion trying to build the X-30, a single-stage spaceship powered by scramjets (with help from rockets, of course). It never flew. At the same time, for comparison, NASA's Langley Research Center studied building a single-stage pure-rocket spaceship. The results were interesting.
The pure-rocket design was more than twice as heavy as X-30 at takeoff, because of all that LOX. On the other hand, its empty weight - the part you have to build and maintain - was 40% less than X-30's. It was about half the size. Its fuel and oxidiser together cost less than half as much per flight as X-30's fuel. And finally, because it quickly climbed out of the atmosphere and did its accelerating in vacuum, it had to endure rather lower stresses and less than 1% of X-30's friction heating. Which approach would be easier and cheaper to operate was pretty obvious.
The Langley group's conclusion: if you want a spaceship that operates like an aeroplane, power it with rockets and only rockets.
There have been some other discussions of this lately, but I would say the pursuit of SSTO resulted in a lost decade for spaceflight in the 1990s.
SSTO is just barely possible, the problem is that you have a big rocket that carries a tiny payload so you are driven to exotic engines, exotic materials, and various risky technologies.
If Musk had any good idea it was not only falling back to two-stage-to-orbit reusable rockets but also recognizing that it was worth just reusing the first stage. A SSTO gets closer to aircraft-like operations in that you don't need to stack two stages on top of each other, but given how much TSTO improves everything else it's probably worth just optimizing the stacking.
And I strongly suspect Henry knew the "don't turn an airplane into a launcher" extended to using wings for landing and takeoff as well, although in 2009 that maybe wasn't quite as inescapable a conclusion as it is today.
Seems that a judge does not understand the impact of asking company X to "retain all data" and is unwilling to rapidly reconsider. Part of what makes this newsworthy is the impact of the initial ruling.
Retention orders of this kind are not uncommon and the judge has not ordered it be turned over to anyone until they hear arguments on it.
I note with amazement that tons of hn users with zero legal experience, let alone judge experience, are sure its the judge who doesn't understand, not them. Based on what I don't know but they really are sure they get it more than the judge!
Underlying this issue is that the judicial system (or the patent system, or the political system) is not populated with enough individuals possessing software engineering "common sense."
It is highly likely that this is not confined to just software, I'm sure other engineering or complex disciplines feel the same way about their discipline.
How do we have experts inform these decisions without falling into the trap of lobbying where the rich control the political and legal sphere?
Anyway, I cede you the point that the US law does not match my "common sense" esp around this 3rd party rule mentioned in other comments. It kind of sucks that US "winning the internet" means that even non-US citizens are subject to US law in this regard.
The judicial system is supposed to apply the law, not "common sense". How could it be otherwise? If you don't like the law then take that up with the legislative branch.
Who is meant to pay for all this data retention? If OpenAI win the argument, can they claim the storage costs from plaintiffs?
It's OK to say "don't throw out a few pieces of paper for a bit", but that doesn't compare to "please spend $500k/month more on S3 bills until whenever we get around to hearing the rest of the case". (Perhaps that much money isn't that important to either side in this _particular_ case, but there is a cost to all this data retention stuff).
> Seems that a judge does not understand the impact of asking company X to "retain all data"
You can count on the fact that the judge does in fact understand that this is a very routine part of such a process.
It is more like the users of ChatGpt don't understand the implication of giving "the cloud" sensitive information and what can happen to it.
It might surprise many such users the extent that the data they casually place at the hands of giant third parties can be, and has routinely, been the target of successful subpoena.
As an illustration, if two huge companies sue each other, part of the legal process involves disclosure. This means inhaling vast quantities of data from their data stores, their onsite servers, executives laptops. Including those laptops that have Ashley Madison data on them. Of course, part of the legal process is motions to exclude this and that, but that may well be after the data is extracted.
Your two examples don't map to the concern about data privacy.
Speed cameras only operate on public roads. The camera in the store is operated by the store owner. In both cases one of the parties involved in the transaction (driving, purchasing) is involved in enforcement. It is clear in both cases that these measures protect everyone and they have clear limits also.
Better examples would be police searching your home all the time, whenever they want (This maps to device encryption).
Or store owners surveilling competing stores / forcing people to wear cameras 24/7 "to improve the customer experience" (This maps to what Facebook / Google try to do, or what internet wire tapping does).
> searching your home all the time, whenever they want
What? How does OpenAI map to your home at all? This is pure nonsense. You seem to have entirely dismissed the comparison to driving a little too out of hand.
The internet is, like the roads, public infrastructure. You can't claim that encryption makes all traffic on the public infrastructure as private as staying home.
You sound like one of those "free man of the land" guys: "I'm not driving your honor, I was traveling."
Okay, so lets move back to a driving metaphor, since you seem to prefer that.
If you drive in the US, the contents of your car are somewhat protected by law. It's not trivial for authorities to obtain the right to search your car, though I understand it's common for some to flout the rules and search anyway.
But there are not gates on the highway which scan all cars and subject the contents of all the cars along with details of the drivers to government scrutiny. Not just the number plates, everything in and about the car.
However, it is the goal of several agencies to perform this level of searching to all data transmitted. Why do you claim that people have less right to privacy when transmitting data in the virtual vs the physical world?
It's hard to me to find heads or tails in your argument.
The argument you're sketching out, of an authority (the nebulous government, the mysterious agencies) totally surveilling your car, is completely and totally disconnected from the discussion at hand. This is a case about two private entities that that are trying to resolve their dispute in a court of law. There's no government agency involved. Whatever you believe about the scary agencies has no bearing on the outcome of this suit.
You say "likely never" but eventually we'll have covered the earth's surface in solar panels. Unless we are transitioning to space based solar and transmission we'll want fusion to increase energy generation beyond surface irradience of earth.
In the meantime, solar panels for massive generation also incur transmission costs to centralize that energy for any major energy usages. We might want to keep having high power generators next to super-high energy consumers. For instance our (theoretical) hyperspace communication and computation
array. Right now those usages are things like Arc Furnaces, Aluminum smelters, data-centers, ...
Plus, we'll want to have figured out that fusion tech so we can build it into our spaceships travelling out beyond Mars as an energy source and hopefully also a thrust source. We want to master that tech on Earth's surface for sure.
> You say "likely never" but eventually we'll have covered the earth's surface in solar panels. Unless we are transitioning to space based solar and transmission we'll want fusion to increase energy generation beyond surface irradience of earth.
> In the meantime, solar panels for massive generation also incur transmission costs to centralize that energy for any major energy usages. We might want to keep having high power generators next to super-high energy consumers. For instance our (theoretical) hyperspace communication and computation array. Right now those usages are things like Arc Furnaces, Aluminum smelters, data-centers, ...
Well before we cover the entire globe in PV, the mere fact that the panels absorb a lot of light means they will change the planet's albedo, heating things up.
But any source of power on that scale will also increase the planet's equilibrium temperature (regardless of if it's PV, fusion, or even if we figure out how to harness dark energy/zero point shenanigans) so we want space-based power before then — and the industrial capacity to use that power in space, because simply beaming it down to Earth is still going to heat up the planet just like any other power source.
Before we even get to that point (in fact, already today) humanity is manufacturing enough metal to make a global power grid with only 1 Ω of resistance the long way around. The limiting factor is geopolitical, not technical, because it's literally just China making enough of the relevant metals.
In any discussion of far-future considerations like this we need to remember that thermodynamics requires all energy used for work to become heat. Covering the Earth's surface in solar panels is one thing, but by the time we're covering every inch in fusion power plants, we've turned the Earth's surface to lava from all the waste heat. There's a physical upper bound on energy usage within the Earth's biosphere that prevents useful energy production from scaling to infinity, and as a result we'll find ourselves optimizing for economics rather than being limited by energy production per unit of area.
As for industry, yes, it seems likely that industry will still represent a large and relatively centralized consumer of power which may benefit from a large dedicated installation. But I'm not convinced that fusion will ever be more economical than even traditional nuclear fission energy (just because your fuel is relatively cheap doesn't mean a thing if the plant itself is essentially disposable because of the energies involved).
As for space, maybe, though considerations of space exploration aren't driven by economics, so whether or not anybody ever figures out a viable fusion reactor design for a spaceship could have as little relevance to civilian power generation as your classic RTG did.
This optimization is important because you can then discard most of the number in question, bounding the integer size required for computation.
For instance you could store the number in question in a 128 bit integer, shift left (double), check for odd digits (a series of modulo & divide operations) and then truncate using a modulo and subtract. You can repeat this process as long as you like. If you find an all evens number than you can do a more expensive indepth check.
When someone is talking about insulating shoes their point is that the body is electrically isolated from the floor. Without that isolation charge can travel between the two. Concrete and skin are fairly good conductors by comparison with air or insulators.
Wearing a conducting ring might make it easier for charge from the air to move into your body through your skin - but it will not make it easier for that charge to get to the floor (and then to ground) from your body.
Generally when people talk about shoes (or tires) they are talking about voltages that can jump the distance from their foot to the ground through air - around the shoe.
Most shoes are not great insulators - they insulate but how knows who much. electricians sometimes buy special shows that do insulate. Those shoes come with care instructions and dust on the outside compromises their insulation.
While you raise some valid points, you need 10kV to jump a 1cm gap. So in a domestic situation at the much lower voltage involved (130V / 250V) I imagine you don't need to worry so much about the air gap.
The care instructions and dust you mention sounds likely to be super important when you're casually touching live stuff while standing on a grounded floor.
In an electrostatic situation, the electrons can flow (almost) freely without the shoes / tyres and so a large differential between body / ground will not build up. With shoes, you'll need a large difference to build up (10kV? 20kV? More?) before it discharges.
Anyway, there is clearly a difference between these situations - even if the shoes don't provide magical protection. But the shoes are not magical, as you correctly describe.
The skin is very resistive, the body itself is not. There is a soldier who "won the darwin award" by taking the probes of a multi-meter and after measuring their skin resistance decided to measure their internal resistance.
After piercing the skin, the test current from the multimeter (9v) was sufficient to electrocute this person. Sadly it (apparently) was a fatal injury. I couldn't find a reference, but the logic makes sense (50V sufficient to kill normally, skin is most of the protection).
When dealing with electricity, having items which reduce the protection your skin offers (metal rings, moisturizer, etc) is a substantial risk.
Well, more correctly the difference needs to exceed the breakdown voltage of the insulation barrier. Or (depending on the insulator) some current might flow the entire time, but a limited amount.
If you change 1,2,3,4 into A,B,C,D - you're still counting. You're just using different symbols to do it.
Take a moment longer to think about it and you realize that 1,2,3,4 are also just symbols - potentially raw counting does not require these symbols at all.
So ABCD is unlikely to make life easier for the subject of this story.
We are in agreement then. I'm not saying it would make things easier, it's just convention. Being a lifelong musician, I will admit that it's so internalized in me, I never have to count or even think about numbers in any way, unless I'm counting some foreign complicated time signature. Honestly I think the tabla guys got it right with their syllabic mnemonics.
It also helps you to move away from familiar but hacky solutions towards much more standard and robust ones - the AI doesn't approach the problem with your many years of battle scars and baggage but instead sees your problem space with fresh eyes and recent data.