Promises are cheap with this admin, don't count any money until it's actually being paid out. Used to be I'd say until it's in a bill but this administration claims the unilateral right to cut any funded program.
Prompting with documentation and examples works. In an agentic tool having an MCP server for the db helps assuming it is a straightforward schema with explicitly defined relationships. Also helps if the tables correspond to entities in a natural way.
We're pretty locked down around AI tools. Right now we can only really use GH Copilot, it's been ok so far though it's funny to see it suggest edits then suggest the opposite on the next time it review the PR if you accept them.
We've launched dozens of shuttles below the minimum temp threshold before there's no reason to delay this launch...
Also that's a singular industry, if the current crop of AI companies deliver what their hype and valuation demands it's a shock across the whole economy not isolated.
200 years ago, 95% of the workers in my country worked in subsistence farming. Today, only 2% are farmers. The whole spectrum of labor has turned upside down and upside down again, in that time. It has certainly not been a singular industry.
> 200 years ago, 95% of the workers in my country worked in subsistence farming. Today, only 2% are farmers.
Yes, it took some time to go from manual/animal labor (energy used is food) to mechanical labor (mostly oil energy). And oil is more energy dense than food, and tractors are more powerful than horses. And bonus points for the oil, it allowed to build fertilizers to boost productivity per acre. So, yes eventually we just need 2% to do what what 95% used to do in farming.
AI is promising to do the same but in virtually all industries (manufacturing, services, healthcare, etc.) and in a way shorten span.
Work used to be labor (human/animal) fueled by energy (food) + intelligence (human) fueled by energy (food), then labor (machine) fueled by energy (oil/electricity) + intelligence (AI) fueled by energy (electricity).
IF work is mostly done by AI/machines fueled by energy. Then work's price is mostly a function of energy price (assuming materials can be extracted/transported/transformed is also a function of energy).
If energy becomes abundant and cheap, then there is no reasons to not let AI do the work.
But then what happens to the rest of us, how the economy keeps humming ?
It was a bunch of small changes over the course of 200 years but yes that's perfectly comparible to the effects needed to justify the valuation put into all the AI companies right now... but I was talking about the issues with comparing it to singular inventions like the cotton gin or jacquard loom that DO largely only affect one industry.
I think it's weird there's so much pushback on the idea that if the hype proves true and it /can/ replace basically any knowledge worker (and potentially drive robots replacing physical laborers) that that would have a bit of a larger effect than inventions that affect some parts of some industries...
There's plenty of space to think it just won't happen (where I'm personally at, at least on the current LLM driven versions) but if it does work the broad spread of the impact would require a huge amount of change all at once.
> I was talking about the issues with comparing it to singular inventions like the cotton gin or jacquard loom
Ok, appreciate the clarification. But in that time frame there have been a number of really tectonic inventions that changed pretty much everything: steam power, ICE power, electrification, refrigeration, computing and the internet, just to name a few off the top of my head.
> There's plenty of space to think it just won't happen (where I'm personally at, at least on the current LLM driven versions)
Same. I am both optimistic about human ability to find new jobs, and skeptical that "AI" is going to make that necessary in the new future.
That amount of change over 200 years is vastly different from the supposed timeline for AI to 'change everything' is the core of the difference. Over that long there's time for people to retrain into other jobs and there's enough people not significantly affected by the change that society as a whole can roll on and support the affected people. Mass disruption and joblessness is extremely destabilizing.
I get what you are saying, and I don't think you are wrong, but it has been like 100 monumental changes in 200 years. The demographic shift of the industrial revolution was particularly painful, to be sure. But we seem to be pretty good at coping with them, overall.
And again, I remain skeptical that general artificial intelligence is actually that close at hand.
There is likewise no indication that it won't. And if I am looking at a pattern where a thousand careers were destroyed by the advance of technology and were swiftly replaced by tens of thousands of new ones, it is not unreasonable to suspect that the pattern is likely repeat.
Do you think that the velocity of change is different from previously ?
I am wondering if we are touching on a human biological limitation. Human are adaptable and flexible, but there is a limit to that flexibility. Some sort of biological limit on how fast we can turn around.
The technology acceleration is increasing, and I am wondering if there would be a point where the technology would evolve faster than what human biology can comprehend.
1,000 years ago, anyone could pretty much build or fix the current technology (anyone could fix a cart).
50 years ago, a majority of people could build or fix the current technology (e.g. most could fix a car).
this year, a limited number of people can build or fix the current technology (e.g. how many people can fix a self driving car?)
10 years from now, a very limited number of people if any could build or fix the current technology (e.g. explain how is AI doing this thing?)
If AI evolves at the same pace, and replacing labor (robots) and services (AI), I am not sure that human would turn around? How do you think we can turn things around ?
Education ? but we are reaching the limit already of how much technology we can teach in a student lifetime. Now we could argue, that one does not need a PhD in computer science to use AI, but eventually do we even need someone to use AI ? Would AI be cheap and pervasive enough that AI would drive AI would drive AI... why would you add a 20W analog brain in the loop ?
What activity would require human involvement ? Genuinely curious how the technology acceleration in general and AI in particular would affect the economy.
> If AI evolves at the same pace, and replacing labor (robots) and services (AI), I am not sure that human would turn around? How do you think we can turn things around ?
I see no indication that we are close to building a GAI, or that we are close to solving the hallucination problems that severely limit the utility LLMs without human managers. We don't understand how our own intelligence works, or even an ant's. The notion the we are close to replicating or exceeding it seems far fetched to me.
> What activity would require human involvement ?
Nurses, bar tenders, barbers... Hasn't anyone read Player Piano? :)
> How do you think we can turn things around ?
I dunno. Did anyone know how dangerous fire or deadly spear points world work out?
I think that's more than likely just a fig leaf proposed for and by the people that planned to try to only have to by ~50% of the land they actually wanted to enclose. So much of this land is too far away from anything for much meaningful development.
I'm glad the court case came down on the reasonable side that you can't effectively by half the land (as you buy more and more land of course) to gain control of the entire enclosed area which is clearly what all those land owners thought they could achieve. I wonder if they're going to actually buy the enclosed areas now?
> Isn’t that the entire point of government ID of any variety?
Ideally this could be done without deanonymizing accounts to service providers unless the user wants to for a 'verified' account linked to their identity publically but I don't think any digital ID system has been built that way. Imagine it acting like OAuth but instead of passing back an identity token it's just verification of age, platforms would store that which would show they had performed the age verification and could be used for other age gates if there are any.
That's how EU's digital wallet is supposed to work:
> The selective disclosure of attributes will allow you to only share the specific information requested by a service provider, without revealing extra information.
> For example, with the selective disclosure of attributes you could choose to share your date of birth, but without revealing any other identifying details that could be used for profiling.
You're totally right that it would be easy from a tech perspective to do that. it's a shame that:
(A) most people cannot grasp how it could be that "GovSSO" can attest "This person you just sent our way just logged into GovSSO [with biometric 2FA], and they are at least 16 years old" without the receiving system having any way of knowing who that citizen is or even whether they're 16 or 99.
(B) very real terrible government policies the UK has (like jailing people for speech, and like demanding encryption backdoors that compromise the security, at minimum, of the whole of every British citizen's devices, and at worst every device in the world) incline anyone who's paying attention to assume that the government will somehow use anything related to "ID" and "internet" to do idiotic things like figuring out who owns a Twitter account that committed some wrongspeak so the bobbies can come round them up.
> (A) most people cannot grasp how it could be that "GovSSO" can attest "This person you just sent our way just logged into GovSSO [with biometric 2FA], and they are at least 16 years old" without the receiving system having any way of knowing who that citizen is or even whether they're 16 or 99.
The loophole that every kid everywhere would instantly figure out is that they just need to borrow their mom’s ID, their older brother’s ID, or a pay some Internet service $1 to use their ID.
This is why the services aren’t designed to totally separate the ID from the account. If nothing actually links the ID to the account then there is no disincentive for people to share their IDs or sell their use for a small fee. Stolen IDs would get farmed for logins.
So the systems invariably get some form of connection to the ID itself. The people making these laws aren’t concerned about privacy aspects. They want maximum enforcement of their goals.
> The loophole that every kid everywhere would instantly figure out is that they just need to borrow their mom’s ID, their older brother’s ID, or a pay some Internet service $1 to use their ID.
Do most kids have their parents' ATM card and PIN? Their Gmail credentials and 2FA device? Tons of stuff today relies on a secret the parents aren't supposed to share with their kids. When logging in on a device that wasn't marked "remember this next time" it should be requiring 2FA. Yes, your 19 year old bro can get you porn, but that's been true for like 60 years buying Penthouse at the liquor store.
Of course all this is academic, since the fact is that because things like oAuth are not intuitively grokkable by non-computer people, so no one would accept "having to sign into <porn site> with GovSSO" even if everything was verifiably privacy-respecting.
Coal power is only about 20% of the US power generation at this point so it's not all or even most of a hypothetical EV's power source. So even where it is used 30 MPG is still actually pretty good gas mileage as well especially for their weight and size.
So far, but it is hard to find good EVs on the used market. Sometimes you can, but at the moment I'm looking and there just are not many options at all.
Is it? Will I be able to get parts for it 15 years from now? Will I be able to repair it myself? (I have rebuilt engines before, and I'm planning to replace the transmission in my truck myself this year, so DIY ability is important to me). There have been headlines recently about how unreliable Tesla is.
Transmissions on EVs are generally single stage speed reduction so yes in the extremely rare chance you need to repair it you can. The motors are way harder to repair but they're similarly way less likely to need repair. Other part availability will vary by brand just like with ICE cars/trucks there's nothing magically less available if we're looking at new parts, older common parts there's a bit of a difference just from the designs being newer so there's a smaller population of junked cars to pull random EV pumps off of the way you can with some old car parts.
It's the most popular vehicle in the world by most metrics, so I don't think you need to be concerned about parts. Especially if comparing to any other EV available in the US. Also extremely reliable. Don't trust clickbait headlines, look at consumer reports and satisfaction.
Might be location dependent I had a lot of options last time I was poking around when my current gas car started making potentially expensive sounding noises earlier this year. (they stopped so I'm fine again /s)
It may look better but it's harder to read basically across the board for anyone with difficulty distinguishing letters. Sans serif fonts are easier for people with dyslexia without going all the way to a dyslexia specific font. They're also generally far better for people with all sorts of poor vision.
It really comes down to the fact that it's better to be functional, forms don't need to /look/ good they need to work well. For aesthetic things we can still use the pretty fonts.
For aesthetic or other preferences you change the default font to whatever you please. The default font shouldn't be about aesthetics, it should be first and foremost about usability. Especially on printed media since there it cannot be changed in a whim.
A couple of years ago I went into archives of Dutch newspapers to learn whether and how the famine of hunger in Ukraine (known as Holodomor) was reported back in 1930's. Fuck me, it was hard to read those excerpts. But it is what it is. OCR could've converted the font. The problem is, is the OCR accurate? Like, is my search with keywords having a good SnR, or am I missing out on evidence?
Personally, Times New Roman was likely the reason I did not like Mozilla Thunderbird. I have to look into that.
> The default font shouldn't be about aesthetics, it should be first and foremost about usability.
The thing about usability is that it's both objective and subjective, and one can argue that aesthetics is part of usability. For example, I find writing code much more pleasant with Comic Code font, and I can imagine that there are other people that would hate it.
Sure but I think we could agree it looking nice ranks lower than being structurally more difficult to read for people? If there were a freely preinstalled option that was both sure but given the choice between functional and aesthetic readability wins hands down.
Off topic but did you find anything interesting? I spent a few days researching Holodomor and was surprised how poorly understood it still is even today, and badly reported at the time. Good propaganda case study. There’s a dramatic film about the reporting too, Mr. Jones (2019).
I haven't researched it explicitly, but I do come across "what happens in the wider world" notices in small historical newspapers and sometimes I search to see what it was about. Saw a mention about some general winning an important victory, searched his name, found out he was one of the whites, and the first thing claimed about him was that he only came in "once the war was already lost".
What I found was that yes, it was reported about, but very little. The notable person who did research the event, Gareth Jones, is indeed an interesting story (he was also referenced to by the newspapers). I believe it was underreported, but we could've known. Helped, now that is a different question I don't dare to answer. The Soviets used disgusting tactics in Eastern Europe, see the book Bloodlands.
Printed document isn't what I was on about. There the default should 100% be about accessibility (and then we just want that by default cause we're used to it).
PDF -> Nope.
.doc(x) -> Sure.
Website, OS, apps (including terminal) -> Sure.
Now regarding PDF I might've tried a long time ago when reading some old document (like CIA about MKULTRA). I don't remember if I succeeded. But there are PDF editors out there. I do think it likely screws layout (esp. larger documents), but that can be true for .doc(x) as well.
I think it would be a smaller issue if it only applied to digital media. Presumably though this applies to all media.
And I can certainly confirm that changing the font of PDF will almost always result in a unreadable mess. Something about how a PDF doesn't have text "blocks" and instead fixes each character making text reflow almost impossible.
reply