I imagined one of those automatic lawn mowers on a very sunny day, going about its own business cutting the grass. Until it arrived on the non-shielded side, and it kinda sags a bit and keels over into the black-charred ground on that side. A clean shadow between the un-touched grass, and the ground in directly in the path of the blast. Or maybe the charred ground is for later. We need to see the nice, automated kitchen first. And only then the rest of the world. Would be a very interesting short film, I think.
I have counter evidence of this. My wife with absolutely zero skills in computers was at the terminal doing things that Claude was telling her to do and generated some impressive tools. One was a tool to help her organize her day. It involved scripts, PDF generation, printing. She did it all without even asking me. Maybe it fails in some cases but I'm not sure the anecdote above is the average experience.
Users know to expect different things at different speeds, and have come to expect the actual "buy" button to be slow. 200ms is pretty fast for that, tbf
Who made his guitar? It seems like he made all his art on the back of someone else’s work… they should be getting royalties for every track made with their product
(I’m pointing out the absurdity of trying to assign ownership within a system of shared creation)
DNS level is more secure but less effective because apps and website can serve the essential content on the same domain as the advertisement. It will also become less effective over time.
YouTube does this, so you need to use a browser-based ad blocker.
This is a bit of a doomer article, but quite honestly, 200 billion dollars a year on a 30 billion dollar a year business that is growing does not really sound as bad as the author makes it out to be, especially when that business consists of growth startups that are currently primarily concerned with completely automating your current revenue stream.
The obvious way to recoup spend is to grow, rugpull by cranking up costs 6x and reducing inference cost by half (highly achievable with improved silicon and technology), then simply fire a large percentage of software engineers. From that perspective the current behavior is a bit wicked but downright logical.
And having two whale customers is not really out of the ordinary for any software company... it's just the scale that is staggering.
The actual risk that the author does not even broach upon for investors... the thing that will actually torpedo this massive investment are the open source open weight chinese models that commoditize the entire endeavor. If you don't have a monopoly, you cannot rugpull and 6x the costs on the consumer.
Ironically, the actual thing that will likely kill OpenAI is ACTUAL OPEN AI.
Tangential/meta: Holy shit, I've never seen a thread where almost half the comments are dead (and LLM written), especially for a post that's (currently) at 86 points and 20 comments (4x ratio is "pretty good quality" post signal generally for me).
correct, moderators in that context is dang/tomhow, not just users with downvoting capabilities. user flags will always have [flagged], otherwise downvoting just makes the text gray
Coming from another experienced dev who was frustrated (early) - one tip. Don't over think it. I was tripping up in two system design interviews that ultimately boiled down to them wanting laughably basic answers (one pagination, the other a db transaction). Don't be afraid to deal in basics and then directly ask if they'd like something more detailed, or not. Many places, especially non-elite ones, treat system design as a _very_ basic test of knowing how pieces fit together.
The first computer I used (Sinclair Mk14) featured two fuse programmable ROMs. Fwiw the article says PROMs were replaced by EPROMs in the 1980s which is true but only for memory applications. PALs (programmable array logic, also made by MMI), which were the bread and butter of random logic design through the 1990s afaik also used fuse programmability. Eventually those were replaced by more dense CMOS chips: FPGAs and so on that use either EEPROM or RAM programming. There is also a factory programming technique used on large chips where fuses are "laser zapped" to lock/unlock functionality, set limits on clock speed and seal encryption keys, which I think is still in use today.
It makes breaking and entering legal because no breaking to enter happens. It allows people to move in and squatting is now in play so they could lock you out and call police if you try to enter.
I might be dense but from what I read I have no idea why any of the stolen documents have any value to OpenAI in the context of their business. Knowing Apples maybe future products etc seems pointless.
How does your heuristic handle your qualitative and quantitative words here? If you land on a blog post with a single, pertinent illustration (from a LLM), do you assume the post, site and author are all robot trash and run away? Investigate the other posts before reading further?
The point of my caps, above was that I'm writing for me. If I think my post is improved by me prompting a LLM for an illustration, that's up to me. I'm not writing for your heuristic, I'm trying to express myself every way I can. The time where you find individuals untainted or unaided by these tools is over.
I'd try to just content by it's content rather than it's illustrations.
SOC 2 is almost a must-have for startups if they want to pursue enterprise deals. Not all soc 2 audits worth the same. People usually trust 2 types of audits:
1. a reputable name or
2. a transparent report that can show all the details like what evidence the auditor checks, how the auditor checks it, and what the auditor finds.
And soc 2 currently is the mostly widely accepted one. There are other more specialized ones for sure but not as universally accepted as soc2.
> Who writes computer programs that deal with pre-1582 _dates_ at all?
People writing knowledge base software, like Wikidata. It's helpful to be able to display historical dates in any desired calendar, and also to be able to do date arithmetic between any two arbitrary dates.
Yeah Apple does have a history of doing these things. There was Palm, where Jobs told them if they didn't stop recruiting Apple employees he would destroy Palm with patent lawsuits. Jobs was also one of the main architects behind the no poach cartel in SV between Google, Adobe, and Intel, and Apple.
More recently there was Nuvia, which the lawsuit was dropped, and Rivos, which was settled.
> An infinite/unbounded universe can literally be modeled by a single random infinite sequence, though. Even a random binary sequence. Think about it: Every logical structure, every pattern, must recur infinitely many times. A complete description of our universe, also, will be among them.
Your understanding is wrong; That is not how Randomness and Infinity works. I used specific qualifiers "true uncaused nondeterministic" for Randomness which mean that there is no underlying process, no known deterministic causes, no known sequence/list, absolute unpredictability etc. Now remember the difference between "Countable Infinity" and "Uncountable Infinity" and how in the latter case no list can ever capture every element in those sets (see Cantor's diagonalization argument). So when truly nondeterministic randomness operates over an uncountable sample space, your argument does not hold.
For a concrete example, think about the "Copenhagen Interpretation" of Randomness and the Uncertainty Principle in Quantum Mechanics.
> Consider this: Let's say you have a physical model of the solar system and you want to specify the position of Venus to fantastically high accuracy, absolute or near-absolute precision. First, you'd have to take into account the position of every body in the solar system, then you would need to fold in the influence of nearby stars, then the whole galaxy, then the full stress-energy content of the observable universe -- and beyond that our notions of "position" themselves break down. In our universe, everything is in principle simulatable, but no finite agent can ever measure or compute to arbitrary precision ... So you can't say "arbitrary precision can't be measured, thus the universe is unsimulatable in principle" because the first statement there is trivial and the second doesn't really follow and is still far out of reach.
Again No; "In our universe, everything is in principle simulatable" is false since you cannot know every single parameter affecting the object under study. Precision is a consequence of that. There are parameters which we do not know and cannot know since the Universe is not deterministic at atomic scale (QM again). This fundamental probabilistic indeterminacy underpins everything.
In this context you should also look into "Complex Systems Science"(CSS) where components self-organize into whole systems with unpredictable behaviour and emergent properties none of which can be known beforehand from the components.
> Consciousness is the most common objection, and yet there's absolutely nothing to suggest that it isn't simulatable. Hence the enduring popularity of "the universe isn't real and we're post-singularity ancestor simulation" theories, and similar. They're popular precisely because there's nothing to disprove them, and it's not even clear that they can be disproved.
Nope. One model of "Consciousness" is as an emergent property where biological/chemical/electrical/etc. systems come together in a Complex System eg. an Organism i.e."Us". This is the materialistic scientific view. It is unique to that instance of organism and cannot be simulated in-toto since it comes into existence only in the whole organism and nowhere else.
Finally; i want to point out that while a Mathematical Model can approximate Reality, a Computer Model is an approximation of the Mathematical Model and hence twice removed from Reality.
I assume there is a way to handle this in modern contract law. Surely people have forgotten how long the various months are and written contracts referring to April 31.
