This seems like a good set of ideas if you can guarantee that you'll never have false alarms. I've had too many birds in warehouses and employees forgetting their codes to feel comfortable going full-hell-interior on alarm.
Birds? Yeah, perhaps. We didn't have any trouble with false alarms that I recall at one shop I worked at with a (relatively small, alarmed) warehouse space where the overhead door was usually open during warm days. I can see it happening, but the false alarms would happen regardless of the intensity of interior alarms.
And the system should not be armed when desirable people are inside, so that problem seems like it is for the birds.
When employees forget their codes and trip the alarm when they're the first ones into the shop at whatever time, they can just go outside to escape the hellish indoor torment. Not perfect, but not so bad either when the goal is to keep people out. :)
Perhaps the smoke should have a harder trigger than the noise, though, if for no other reason than it's a consumable that eventually needs to be fed more money every time it is activated.
I'm also on the $100 max plan. I let Fable rip on a complicated issue involving hot-reloading modules in a GUI app built with Racket, it's fixed a couple issues over the last hour, and I've used about 17% of my session (not weekly) limit.
Yes, which goes in line with the argument that claiming that it's "the most deployed" as proof of superiority or suitability for any use case is equivalent to claiming the same for Internet Explorer. It's the most deployed because it's bundled in a lot of systems, not because people are purposefully using it as a DBMS.
But it doesn't, because none of those systems are presenting SQLite to the user as something they should be using; they don't even make SQLite available to the user at all. Those systems all use SQLite internally to manage data.
> My technopolitics faction – the faction associated with the Electronic Frontier Foundation, where I've worked for a quarter-century – has an answer: the role of encryption is to provide a measure of privacy and security that is best used to organize political struggles to demand the rule of law and respect for human rights. Encryption isn't proof against rubber hoses, but it is effective against many other forms of state repression, and it can provide a technical edge for those engaged in a political struggle.
> Another faction – the faction most associated with bitcoin and subsequent cryptocurrency projects – rejects the role of the state altogether, and seeks to replace states (and state-regulated institutions like courts and banks) with mathematics. Rather than asking courts to interpret contracts, we can put our trust in self-executing "smart contracts," and rather than asking banks to safeguard our financial integrity, we can use cryptographic software to ensure that money only moves when the person it belongs to tells it to.
So he's saying there is a split between those who believe the state and the rule of law are essential tools of freedom, and those who believe technology can provide its own law and guarantees without any need for the state. None of that is incompatible with the EFF being a libertarian project.
And your confusion derives from…what? When he explains this, you feel the correct response is basically "nuh-uh"?
David Gelernter describes a theory of consciousness and creativity that explains why this works in his book “The Muse in the Machine”. I recommend it to everyone.
This sounds like you don’t have much exposure to actual professional engineering disciplines. I’m sure civil, electrical, structural and mechanical PEs would be quite surprised to hear there are no guardrails on their professions.
I have fond memories of a boss who was an actual, licensed engineer while the rest of us were very much normal software devs. Boss was pretty chill except when someone someone suggested we should be called "engineers" rather than "developers", at which point they said "if you guys were building bridges, people would be dead." (I don't think all software needs to be built to rigorous engineering standards but man... I think about that line a lot.)
I prefer the term "software developer" and that's what I use when I don't need the prestige of the term "software engineer". It's disadvantageous for organizations to do that with actual job titles, though.
Absent US government intervention to codify the term "engineer", probably the only way out of the "engineer" trap is through further title inflation, where the developers all become "vice presidents". :)
Yeah, it's 100% the better term. We've got rules against using engineer here in Canada though several companies I've worked for have called me an engineer. Apparently Professional Engineers Ontario sometimes goes after people for calling themselves engineers but I've never heard of it actually happening, and I don't know that they have any real teeth given that the places I worked that called me an engineer were Canadian-owned. (In fact, the only place where they checked if I could use the title was the one multi-national. Go figure.)
In my experience “engineers” and builders are often quite “conservative” and really don’t like pushing the envelope, and they often only do it under protest.
The most famous example may be the perpetual war between architects and engineers/builders.
Researchers need to go wild and sometimes far off-the-rails to increase the odds of coming up with something that is both new, and potentially popular enough, if they want the option to attract marketers who can only thrive on mass-consumption.
With luck, one out of 100 inventions will show promise on those points.
There's always a lifetime wake where the overwhelming vast majority of the work remains undeployed no matter what. The more undeployed milestones and inventions that some scientists have under their belt, the more accomplished they often are whether anybody knows it or not.
OTOH, equally active engineers more often need to have most of their time engaged in actual deployment of some kind or another, otherwise not as much progress will be able to reach as many people that could benefit. So many times nothing would be accomplished without a long-term focused engineering effort once an objective has been identified. But it can be hard to stop a train when it's already coming off the drawing board at full steam.
It does seem a lot more likely for a judicious researcher to cast off some major progress in what could very well turn out to be an unsavory development, such as likely misuse, even if it could be marketed as the most popular thing they have so far. Just add it to the pile of other things that best remain undeployed. There's plenty more where that came from, and the best is yet to come.
Perhaps popularity alone is not always the best measure of progress.
I actually originally wrote "technologists" but thought that the word sounded kind of odd. Now I realize it better captures what I was trying to convey.
I like the example of the actors' unions in the 1960s, where instead of "fighting" television in the sense of demanding people stop using it, they fought by organizing to get ongoing residual payments whenever their work was repurposed for the new medium. You don't have to stop fighting, you just need to recognize what the real problem is.
The only entities that would make meaningful money from an ai version of this would be IP giants like Disney. Your average guy is not going to get rich off his microscopic amount of data used. Basically Spotify.
I think you read the analogy too narrowly. I too doubt whether micropayments are worth fighting for, but there are other outcomes for which we could and should work together. For example, data center effects on water and power usage are well-known negative externalities of AI industry that could be eliminated by requiring data centers to invest in mitigations. The government could buy large holdings of stock in AI companies and distribute dividends, just like the Alaska Permanent Fund. etc. etc. You can quibble with individual examples here, but the larger point is that there are productive ways of tackling this transition, old man yells at cloud is not one of them
Ok i actually agree with that. Imo the gov is going to have to print a lot of money to deal with the impacts of ai (covid was $5T), if we do it now that same dollar would go 10x further and the people would effectively be the largest owner of AI labs like Anthropic.
An ounce of prevention is worth a pound of cure idea.
The people who hate "AI" are correctly understanding it as a political project, not simply a technology. Ali Alkhatib's definition here is clarifying in this regard: https://ali-alkhatib.com/blog/defining-ai
This was clarifying? It reads like a sleepy undergrad's first attempt, complete with the constant meandering to satisfy some word count. The irony is a SOTA AI could make this person's case far more succinctly and convincingly. You really need to hold yourself (and the people you read) to a higher standard.
This entire brain dump of a blog post could be summed up in one famous sentence: Man is a political animal.
I never understand people who seem to have a need to grasp at such poorly written blogs for an understanding of today's affairs. Humans have really been remarkably consistent in their nature. The answer to your question has already been written, maybe even centuries ago by someone who thought about this a lot harder than you. Sometimes it feels like LLMs are so good simply because most people are far less interesting than they think they are. At some level humanity has been asking the same fundamental questions since the dawn of civilization. At a certain point what more does the average person have to say that we haven't already heard before?
To me this just muddies the waters further. If I run a model on my own hardware am I working with the "AI" political project?
I would agree that there is a political project happening in the AI space (and that it predates modern AI); I think it's worth giving that political project a distinct name, rather than conflating a term already widely used and understood very differently by normal people.
> If I run a model on my own hardware am I working with the "AI" political project?
No. Obviously, what you do on your PC is inconsequential to the rest of society. But despite this, AI and its consequences in big tech have become so thoroughly linked because the entities that develop and profit off of AI use are so big and influential over the rest of us. The hobbyist space isn't what people even think about.
Is it really that common? I know it exists, but everywhere I go I hear of everyone paying for enterprise subscriptions, the implications that LLM usage rates have for businesses, companies burning through allotted tokens, the use of cutting-edge models. From my impression, local models (especially so for image generators) were the domain of the hobbyists and only a small slice of corporate use.
Exactly no one hates the ai algorithm finding cancer, or predicting protein folding or finding novel compounds for pharmaceuticals what they hate is the "we will slurp up all human knowledge violate everyone copyrights and give nothing back in return and get you fired and replaced by a shitty ai chatbot"
This is funny and a pretty clever move, but not actually the argument I'm making. I'm specifically saying you can't make people un-learn math once it turns out to have interesting uses.
That invokes both learning and interest, and the latter can be rolled back. You can't (usually) remove some item from the store of human knowledge, but the humans can lose interest in the item. Interesting uses can cease to be interesting. Fads can pass, you don't know. Don't see the Mandelbrot set around much these days.
OK, but we're talking about people who are "against AI". Are you saying that opposition to AI might help people lose interest in it? I'm not aware of an example of opposition to a useful application of math that caused people to lose interest. It didn't happen with public key encryption, for example. Can you explain further how you see "hating AI" (in the sense of TFA) will cause a loss of interest?
> Don't see the Mandelbrot set around much these days.
Was computing the mandlebrot set ever shown to be broadly and commercially useful in some way?
The problem is that isn't just math, the problem is more specifically one of the weakest maths for our entire species: Probability and Statistics. Humans as a general rule are terrible at Probability and Statistics.
Being angry at LLMs isn't hating math, it is hating slot machines and the human inclination to lose everything to them because the Gambler's Fallacy affects us all, unfortunately deeply.
It's math that requires an obscene amount of compute. If it's possible to make DRM chips that don't let you play pirated movies and GPS chips that shut down when going too fast, then I reckon it's also possible to make GPUs that shut down when they encounter anything that looks like a transformer. The problem is regulatory, not technical.
It doesn't require obscene compute though. I can run a model on my macbook with 48GB of RAM that is roughly comparable to Sonnet 4.6. A year from now the same machine will be able to run much more capable models.
I would agree there are sound regulations needed, but banning certain kinds of math is not it. (Your DRM example is particularly unfriendly to your point in this regard.)
chips limiting what kind of software can be run sounds absolutely awful and would be a huge overstep in government power. once I buy a piece of hardware I should be able to run whatever I want on it.
Sure, until it gets to the point where you're no longer able to just "buy a piece of hardware". Then it'll suddenly sound like a pretty sweet deal. Like when Nvidia nerfed ETH mining back when that was still a thing. That too was done to give more people access to hardware, not to take away their freedom.
reply